Sean Carroll: General Relativity, Quantum Mechanics, Black Holes & Aliens

00:00:00.000 | the whole point of relativity is to say there's no such thing as right now when you're far away.

00:00:04.480 | And that is doubly true for what's inside a black hole. And you might think, well,

00:00:09.200 | the galaxy is very big. It's really not. It's some tens of thousands of light years across

00:00:14.960 | and billions of years old. So, you don't need to move at a high fraction of the speed of light to

00:00:21.920 | fill the galaxy. The number of worlds is very, very, very, very big. Where do those worlds fit?

00:00:31.760 | Where do they go? So, the short answer is the worlds don't exist in space.

00:00:40.000 | Space exists separately in each world.

00:00:47.920 | The following is a conversation with Sean Carroll, his third time in this podcast. He is a theoretical

00:00:54.720 | physicist at John Hopkins, host of the Mindscape podcast that I personally love and highly

00:01:00.480 | recommend, and author of many books, including the most recent book series called The Biggest Ideas

00:01:06.320 | in the Universe, the first book of which is titled Space, Time, and Motion, and it's on the topic of

00:01:12.320 | general relativity. And the second, coming out on May 14th, so you should definitely pre-order it,

00:01:18.880 | is titled Quanta and Fields, and that one is on the topic of quantum mechanics.

00:01:24.000 | Sean is a legit active theoretical physicist, and at the same time, is one of the greatest

00:01:30.800 | communicators of physics ever. I highly encourage you listen to his podcast, read his books,

00:01:37.360 | and pre-order the new book to support his work. This was, as always, a big honor and a pleasure

00:01:44.000 | for me. This is Alex Friedman Podcast. To support it, please check out our sponsors

00:01:48.960 | in the description. And now, dear friends, here's Sean Carroll. In book one of the series,

00:01:56.560 | The Biggest Ideas in the Universe, called Space, Time, Motion, you take on classical mechanics,

00:02:02.400 | general relativity, by taking on the main equation of general relativity and making it

00:02:08.240 | accessible, easy to understand. So maybe at the high level, what is general relativity?

00:02:15.440 | What's a good way to start to try to explain it? - Probably the best way to start to try to explain

00:02:20.320 | it is special relativity, which came first, 1905. It was the culmination, right, of many decades of

00:02:27.600 | people putting things together, but it was Einstein in 1905. In fact, it wasn't even Einstein. I should

00:02:32.880 | give more credit to Minkowski in 1907. So Einstein in 1905 figured out that you could get rid of the

00:02:40.640 | ether, the idea of a rest frame for the universe, and all the equations of physics would make sense

00:02:46.800 | with the speed of light being a maximum. But then it was Minkowski, who used to be Einstein's

00:02:52.320 | professor in 1907, who realized the most elegant way of thinking about this idea of Einstein's was

00:02:58.560 | to blend space and time together into space-time, to really imagine that there is no hard and fast

00:03:06.480 | division of the four-dimensional world in which we live into space and time separately. Einstein was

00:03:12.560 | at first dismissive of this. He thought it was just like, oh, the mathematicians are over-formalizing

00:03:18.160 | again. But then he later realized that if space-time is a thing, it can have properties.

00:03:25.440 | And in particular, it can have a geometry. It can be curved from place to place. And that was what

00:03:30.480 | let him solve the problem of gravity. He had previously been trying to fit in what we knew

00:03:36.560 | about gravity from Newtonian mechanics, the inverse square law of gravity, to his new relativistic

00:03:42.960 | theory. It didn't work. So the final leap was to say gravity is the curvature of space-time.

00:03:50.160 | And that statement is basically general relativity. >> And the tension with Minkowski was,

00:03:55.440 | he was a mathematician. >> Yes.

00:03:56.800 | >> So there's a tension between physics and mathematics. In fact, in your lecture about

00:04:01.680 | this equation, one of them, you say that Einstein is a better physicist than he gets credit for.

00:04:09.280 | >> Yep. >> I know, that's hard. That's a little

00:04:13.200 | bit of a joke there, right? >> Yeah.

00:04:14.400 | >> Because we all give Einstein a lot of credit. But then we also, partly based on fact, but partly

00:04:21.120 | to make ourselves feel better, tell ourselves a story about how later in life, Einstein couldn't

00:04:25.920 | keep up. There were younger people doing quantum mechanics and quantum field theory and particle

00:04:30.720 | physics. And he was just sort of unable to really philosophically get over his objections to that.

00:04:36.800 | And I think that that story about the latter part is completely wrong, almost 180 degrees wrong. I

00:04:43.360 | think that Einstein understood quantum mechanics as well as anyone, at least up through the 1930s.

00:04:49.760 | I think that his philosophical objections to it are correct. So he should actually have been

00:04:55.520 | taken much more seriously about that. And what he did, what he achieved in trying to think these

00:05:02.080 | problems through is to really basically understand the idea of quantum entanglement, which is kind of

00:05:08.640 | important these days when it comes to understanding quantum mechanics. Now, it's true that in the 40s

00:05:13.760 | and 50s, he placed his efforts in hopes for unifying electricity and magnetism with gravity

00:05:20.720 | that didn't really work out very well. All of us try things that don't work out. I don't hold

00:05:26.240 | that against him. But in terms of IQ points, in terms of trying to be a clear thinking physicist,

00:05:31.440 | he was really, really great. - What does greatness look like for a

00:05:34.960 | physicist? So how difficult is it to take the leap from special relativity to general relativity?

00:05:42.080 | How difficult is it to imagine that, to consider space-time together and to imagine that there's

00:05:50.400 | a curvature to this whole thing? - Yeah, that's a great question. I think

00:05:55.360 | that if you want to make the case for Einstein's greatness, which is not hard to do, there's two

00:05:59.840 | things you point at. One is in 1905, his famous miracle year, he writes three different papers

00:06:09.200 | on three wildly different subjects, all of which would make you famous just for writing that one

00:06:15.840 | paper. Special relativity is one of them. Brownian motion is another one, which is just the little

00:06:23.120 | vibrations of tiny little dust specks in the air. But who cares about that? What matters is

00:06:28.640 | it proves the existence of atoms. He explains Brownian motion by imagining there are molecules

00:06:32.960 | in the air and deriving their properties. Brilliant. And then he basically starts the

00:06:38.160 | world on the road to quantum mechanics with his paper on, which again is given a boring

00:06:43.280 | label of the photoelectric effect. What it really was is he invented photons. He showed that light

00:06:49.440 | should be thought of as particles as well as waves. And he did all three of those very different

00:06:54.240 | things in one year. Okay. But the other thing that gets him genius status is like you say,

00:06:59.760 | general relativity. So this takes 10 years from 1905 to 1915. He wasn't only doing general

00:07:05.280 | relativity. He was working on other things. He wrote, he invented a refrigerator. He did various

00:07:09.680 | interesting things. And he wasn't even the only one working on the problem. There were other people

00:07:14.320 | who suggested relativistic theories of gravity, but he really applied himself to it. And I think

00:07:22.560 | as your question suggests, the solution was not a matter of turning a crank. It was something

00:07:30.080 | fundamentally creative. In his own telling of the story, his greatest moment, his happiest moment

00:07:38.240 | was when he realized that if the way that we would say it in modern terms, if you were in a rocket

00:07:43.520 | ship accelerating at one G, at one acceleration due to gravity, if the rocket ship were very quiet,

00:07:50.640 | you wouldn't be able to know the difference between being in a rocket ship and being on the

00:07:54.160 | surface of the earth. Gravity is sort of not detectable or at least not distinguishable

00:08:00.480 | from acceleration. So number one, that's a pretty clever thing to think. But number two, if you or I

00:08:05.680 | had that thought, we would have gone, "Huh, we're pretty clever." He reasons from there to say,

00:08:11.840 | "Okay, if gravity is not detectable, then it can't be like an ordinary force, right? The electromagnetic

00:08:18.480 | force is detectable. We can put charged particles around. Positively charged particles and negatively

00:08:23.200 | charged particles respond differently to an electric field or to a magnetic field." He

00:08:28.320 | realizes that what his thought experiment showed or at least suggested is that gravity isn't like

00:08:33.280 | that. Everything responds in the same way to gravity. How could that be the case? And then

00:08:39.520 | this other leap he makes is, "Oh, it's because it's the curvature of space-time, right?" It's a feature

00:08:44.080 | of space-time. It's not a force on top of it. And the feature that it is is curvature. And then

00:08:49.360 | finally he says, "Okay, clearly I'm going to need the mathematical tools necessary to describe

00:08:55.920 | curvature. I don't know them, so I will learn them." And they didn't have MOOCs or AI helpers

00:09:03.040 | back in those days. He had to sit down and read the math papers and he taught himself differential

00:09:07.200 | geometry and invented general relativity. >> What about the step of including time

00:09:11.920 | as just another dimension, so combining space and time? Is that a simple mathematical leap

00:09:18.720 | as Minkowski suggested? >> It's certainly not simple, actually.

00:09:22.960 | It's a profound insight. That's why I said I think we should give Minkowski more credit than we do.

00:09:30.960 | He's the one who really put the finishing touches on special relativity. Again, many people had

00:09:36.640 | talked about how things change when you move close to the speed of light, what Maxwell's equations

00:09:43.840 | of electromagnetism predict and so forth, what their symmetries are. So people like Lorentz

00:09:48.400 | and Fitzgerald and Poincaré, there's a story that goes there. And in the usual telling, Einstein

00:09:53.920 | sort of puts the capstone on it. He's the one who says, "All of this makes much more sense if there

00:09:59.440 | just is no ether. It is undetectable. We don't know how fast." Everything is relative, thus the

00:10:04.320 | name relativity. But he didn't take the actual final step, which was to realize that the underlying

00:10:10.720 | structure that he had invented is best thought of as unifying space and time together. I honestly

00:10:17.040 | don't know what was going through Minkowski's mind when he thought that. I'm not sure if he was so

00:10:23.040 | mathematically adept that it was just clear to him, or he was really struggling it and he did

00:10:28.880 | trial and error for a while. I'm not sure. >> I mean, do you, for him or for Einstein,

00:10:33.360 | visualize the four-dimensional space, try to play with the idea of time as just another dimension?

00:10:38.000 | >> Oh, yeah, all the time. I mean, we, of course, make our lives easy by ignoring

00:10:42.720 | two of the dimensions of space. So instead of four-dimensional space-time, we just

00:10:47.440 | draw pictures of one dimension of space, one dimension of time, the so-called space-time

00:10:52.240 | diagram. But I mean, maybe this is lurking underneath your question, but even the best

00:10:57.920 | physicists will draw a vertical axis and a horizontal axis, and they'll go space-time.

00:11:04.880 | But deep down, that's wrong because you're sort of preferring one direction of space and one

00:11:11.280 | direction of time, and it's really the whole two-dimensional thing that is space-time.

00:11:16.080 | The more legitimate thing to draw on that picture are rays of light, are light cones.

00:11:22.640 | From every point, there is a fixed direction at which the speed of light would represent,

00:11:28.560 | and that is actually inherent in the structure. The division into space and time is something

00:11:34.320 | that's easy for us human beings. >> What is the difference between space

00:11:38.320 | and time from the perspective of general relativity? >> It's the difference between x and y when you

00:11:43.760 | draw axes on a piece of paper. >> So there's really no difference.

00:11:46.960 | >> There's almost no difference. There's one difference that is kind of important, which is

00:11:51.920 | the following. If you have a curve in space, I'm going to draw it horizontally because that's

00:11:57.200 | usually what we do in space-time diagrams. If you have a curve in space, you've heard the motto

00:12:01.760 | before that the shortest distance between two points is a straight line. If you have a curve

00:12:06.960 | in time, which is, by the way, literally all of our lives, right? We all evolve in time. So you

00:12:12.480 | can start with one event in space-time and another event in space-time. What Minkowski points out

00:12:18.240 | is that the time you measure along your trajectory in the universe is precisely analogous to the

00:12:26.240 | distance you travel on a curve through space. By precisely, I mean it is also true that the actual

00:12:33.520 | distance you travel through depends on your path, right? You can go a straight line, shortest

00:12:38.240 | distance, and curvy line would be longer. The time you measure in space-time, the literal time that

00:12:43.680 | takes off on your clock, also depends on your path. But it depends on it the other way. So that

00:12:49.680 | the longest time between two points is a straight line. And if you zig back and forth in space-time,

00:12:55.760 | you take less and less time to go from point A to point B. >>

00:12:59.400 | How do we make sense of that, the difference between the observed reality and the

00:13:07.520 | objective reality underneath it? Or is objective reality a silly notion,

00:13:12.480 | given general relativity? >> I'm a huge believer in objective

00:13:14.880 | reality. I think that objective reality objectively is real. But I do think that

00:13:20.160 | people kind of are a little overly casual about the relationship between what we observe and

00:13:28.560 | objective reality in the following sense. Of course, in order to explain the world,

00:13:34.720 | our starting point and our ending point is our observations, our experimental input,

00:13:39.600 | the phenomena we experience and see around us in the world. But in between, there's a theory.

00:13:46.480 | There's a mathematical formalization of our ideas about what is going on. And if a theory

00:13:54.720 | fits the data and is very simple and makes sense in its own terms, then we say that the theory is

00:14:01.120 | right. And that means that we should attribute some reality to the entities that play an important

00:14:08.800 | role in that theory, at least provisionally until we come up with a better theory down the road.

00:14:12.720 | >> I think a nice way to test the difference between objective reality and the observed reality

00:14:18.560 | is what happens at the edge of the horizon of a black hole. So technically, as you get closer

00:14:28.000 | to that horizon, time stands still. >> Yes and no. It depends on exactly

00:14:33.200 | how careful we're being. So here is a bunch of things I think are correct.

00:14:39.760 | If you imagine there is a black hole spacetime, so like the whole solution to Einstein's equation,

00:14:46.880 | and you treat you and me as what we call test particles. So we don't have any gravitational

00:14:52.880 | fields ourselves. We just move around in the gravitational field. That's obviously an

00:14:56.320 | approximate approximation, but let's imagine that. And you stand outside the black hole,

00:15:01.760 | and I fall in. And as I'm falling in, I'm waving to you because I'm going into the black hole.

00:15:07.760 | You will see me move more and more slowly. And also the light from me is redshifted,

00:15:14.960 | so I kind of look embarrassed because I'm falling into a black hole. And there is a limit.

00:15:19.920 | There's a last moment that light will be emitted from me, from your perspective, forever. Now,

00:15:27.520 | you don't literally see it because I'm emitting photons more and more slowly because from your

00:15:34.720 | point of view. So it's not like I'm equally bright. I basically fade from view in that picture. So

00:15:41.760 | that's one approximation. The other approximation is I do have a gravitational field of my own,

00:15:47.520 | and therefore, as I approach the black hole, the black hole doesn't just sit there and let me pass

00:15:52.880 | through. It kind of moves out to eat me up because its net energy mass is going to be mine plus its.

00:16:00.960 | But roughly speaking, yes. So I don't like to go to the dramatic extremes because that's where

00:16:05.440 | the approximations break down. But if you see something falling into a black hole,

00:16:08.720 | you see its clock ticking more and more slowly. >> How do we know it fell in?

00:16:13.120 | >> We don't. I mean, how would we? Because it's always possible that right at the last minute,

00:16:19.040 | it had a change of heart and starts accelerating away, right? If you don't see it pass in,

00:16:24.560 | you don't know. And let's point out that as smart as Einstein was, he never figured out

00:16:28.880 | black holes, and he could have. It's kind of embarrassing. It took decades for people

00:16:34.320 | thinking about general relativity to understand that there are such things as black holes.

00:16:39.280 | Because basically, Einstein comes up with general relativity in 1915. Two years later,

00:16:44.960 | Schwarzschild, Carl Schwarzschild, derives the solution to Einstein's equation that represents

00:16:52.640 | a black hole, the Schwarzschild solution. No one recognized it for what it was until the '50s,

00:16:57.920 | David Finkelstein and other people. And that's just one of these examples of physicists not

00:17:02.720 | being as clever as they should have been. >> Well, that's the singularity. That's the

00:17:07.280 | edge of the theory, the limit. So it's understandable that it's difficult to imagine

00:17:12.480 | the limit of things. >> It is absolutely hard

00:17:15.280 | to imagine, and a black hole is very different in many ways from what we're used to. On the other

00:17:19.920 | hand, I mean, the real reason, of course, is that between 1915 and 1955, there's a bunch of other

00:17:26.320 | things that are really interesting going on in physics, all particle physics and quantum field

00:17:30.240 | theory. So many of the greatest minds were focused on that. But still, if the universe hands you a

00:17:35.920 | solution to general relativity in terms of curved spacetime and it's kind of mysterious, certain

00:17:40.960 | features of it, I would put some effort in trying to figure it out. >> So how does a black hole work?

00:17:46.160 | Put yourself in the shoes of Einstein and take general relativity to its natural conclusion

00:17:51.520 | about these massive things. >> It's best to think of a black hole as not an object so much as a

00:17:57.360 | region of spacetime, okay? It's a region with the property, at least in classical general relativity,

00:18:04.240 | quantum mechanics makes everything harder, but let's imagine we're being classical for the moment.

00:18:07.840 | It's a region of spacetime with the property that if you enter, you can't leave.

00:18:12.160 | Literally, the equivalent of escaping a black hole would be moving faster than the speed of light.

00:18:19.040 | They are both precisely equally difficult. You would have to move faster than the speed of light

00:18:22.720 | to escape from the black hole. So once you're in, that's fine. In principle, you don't even notice

00:18:29.120 | when you cross the event horizon, as we call it. The event horizon is that point of no return,

00:18:33.760 | where once you're inside, you can't leave. But meanwhile, the spacetime is sort of collapsing

00:18:39.520 | around you to ultimately a singularity in your future, which means that the gravitational

00:18:46.400 | forces are so strong, they tear your body apart and you will die in a finite amount of time.

00:18:51.600 | The time it takes, if the black hole is about the mass of the sun,

00:18:55.200 | to go from the event horizon to the singularity takes about one millionth of a second.

00:19:00.800 | >> And what happens to you if you fall into the black hole? If we think of an object as

00:19:07.440 | information, that information gets destroyed. >> Well, you've raised a crucially difficult point.

00:19:15.600 | So that's why I keep needing to distinguish between black holes according to Einstein's

00:19:21.440 | theory of general relativity, which is book one of spacetime and geometry, which is perfectly

00:19:26.080 | classical. And then come the 1970s, we start asking about quantum mechanics and what happens

00:19:32.960 | in quantum mechanics. According to classical general relativity, the information that makes

00:19:37.440 | up you when you fall into the black hole is lost to the outside world. It's there, it's inside the

00:19:44.160 | black hole, but we can't get it anymore. In the 1970s, Stephen Hawking comes along and points out

00:19:50.640 | that black holes radiate. They give off photons and other particles to the universe around them,

00:19:56.880 | and as they radiate, they lose mass and eventually they evaporate, they disappear.

00:20:01.280 | So once that happens, I can no longer say the information about you or a book that I threw in

00:20:08.640 | the black hole or whatever is still there, is hidden behind the black hole, because the black

00:20:12.080 | hole's gone away. So either that information is destroyed, like you said, or it is somehow

00:20:18.720 | transferred to the radiation that is coming out, to the Hawking radiation. A large majority of

00:20:24.800 | people who think about this believe that the information is somehow transferred to the

00:20:29.360 | radiation and information is conserved. That is a feature both of general relativity by itself

00:20:36.160 | and of quantum mechanics by itself. So when you put them together, that should still be a feature.

00:20:40.320 | We don't know that for sure. There are people who have doubted it, including Stephen Hawking for a

00:20:44.080 | long time. But that's what most people think. And so what we're trying to do now in a topic which

00:20:51.200 | has generated many, many hundreds of papers, called the Black Hole Information Loss Puzzle,

00:20:56.000 | is figure out how to get the information from you or the book into the radiation that is escaping

00:21:01.840 | the black hole. - Is there any way to observe Hawking radiation to a degree where you can

00:21:08.720 | start getting insight? Or is this all just in the space of theory right now?

00:21:12.080 | - Right now, we are nowhere close to observing Hawking radiation. Here's the sad fact. The

00:21:18.320 | larger the black hole is, the lower its temperature is. So a small black hole, like a microscopically

00:21:26.320 | small black hole, might be very visible. It's given off light. But something like the black

00:21:30.080 | hole at the center of our galaxy, three million times the mass of the Sun or something like that,

00:21:35.200 | Sagittarius A*, that is so cold and low temperature that its radiation will never be observable.

00:21:42.400 | Black holes are hard to make. We don't have any nearby. The ones we have out there in the

00:21:47.440 | universe are very, very faint. So there's no immediate hope for detecting Hawking radiation.

00:21:51.040 | - Allegedly, we don't have any nearby. - As far as we know, we don't have any nearby.

00:21:55.520 | - Could tiny ones be hard to detect somewhere at the edges of the solar system maybe?

00:21:59.520 | - So you don't want them to be too tiny or they're exploding. They're very bright,

00:22:05.360 | and then they will be visible. But there's absolutely a regime where black holes are

00:22:09.200 | large enough not to be visible because the larger ones are fainter, not giving off radiation,

00:22:13.760 | but small enough to not have been detected through their gravitational effect, yeah.

00:22:16.960 | - Psychologically, just emotionally, how do you feel about black holes? Do they scare you?

00:22:21.280 | - I love them. I love black holes. But the universe, weirdly, makes it hard to make a

00:22:26.720 | black hole, right? Because you really need to squeeze an enormous amount of matter and energy

00:22:32.320 | into a very, very small region of space. So we know how to make stellar black holes. A supermassive

00:22:39.520 | star can collapse to make a black hole. We know we also have these supermassive black holes at

00:22:44.720 | the center of galaxies. We're a little unclear where they came from. I mean, maybe stellar black

00:22:50.080 | holes that got together and combined, but that's, you know, one of the exciting things about new

00:22:57.280 | data from the James Webb Space Telescope is that quite large black holes seem to exist relatively

00:23:03.680 | early in the history of the universe. So it was already difficult to figure out where they came

00:23:07.760 | from. Now it's an even tougher puzzle. - So these supermassive black holes are

00:23:12.320 | formed somewhere early on in the universe. I mean, that's a feature, not a bug, right,

00:23:17.120 | that we don't have too many of them. Otherwise we wouldn't have the time or the space to form

00:23:23.680 | the little pockets of complexity that we'll call humans. - I think that's fair, yeah. It's always

00:23:30.160 | interesting when something is difficult, but happens anyway, right? I mean, the probability

00:23:36.320 | of making a black hole could have been zero. It could have been one, but it's this interesting

00:23:40.560 | number in between, which is kind of fun. - Are there more intelligent alien civilization than

00:23:44.800 | there are supermassive black holes? - Yeah, I have no idea, but I think

00:23:50.400 | your intuition is right that it would have been easy for there to be lots of civilizations and

00:23:56.880 | then we would have noticed them already, and we haven't. So absolutely the simplest explanation

00:24:02.240 | for why we haven't is that they're not there. - Yeah, I just think it's so easy to make them,

00:24:07.840 | though. So there must be, I understand that's the simplest explanation, but also--

00:24:12.720 | - How easy is it to make life, or eukaryotic life, or multicellular life?

00:24:16.960 | - It seems like life finds a way. Intelligent alien civilizations,

00:24:21.760 | sure, maybe there is somewhere along that chain a really, really hard leap. But once you start life,

00:24:29.680 | once you get the origin of life, it seems like life just finds a way everywhere,

00:24:34.480 | in every condition. It just figures it out. - I mean, I get it. I get exactly what you're

00:24:39.120 | thinking. I think it's a perfectly reasonable attitude to have before you confront the data.

00:24:45.760 | I would not have expected Earth to be special in any way. I would have expected there to be

00:24:48.960 | plenty of very noticeable extraterrestrial civilizations out there. But even if life

00:24:56.800 | finds a way, even if by everything you say, how long does it take for life to find a way? What

00:25:03.040 | if it typically takes 100 billion years? Then we'd be alone. - So it's a time thing. So to you,

00:25:09.680 | really, there's most likely there's no alien civilizations out there. I just, I can't see it.

00:25:15.360 | I believe there's a ton of them, and there's another explanation why we can't see them.

00:25:19.360 | - I don't believe that very strongly. Look, I'm not going to place a lot of bets here. I would

00:25:23.920 | not, I'm both pretty up in the air about whether or not life itself is all over the place. It's

00:25:29.840 | possible when we visit other worlds, other solar systems, there's very tiny microscopic life,

00:25:36.800 | ubiquitous, but none of it has reached some complex form. It's also possible there's just,

00:25:42.320 | there isn't any. It's also possible that there are intelligent civilizations that have better

00:25:47.360 | things to do than knock on our doors. So I think we should be very humble about these things we

00:25:52.000 | know so little about. - And it's also possible there's a great filter where there's something

00:25:56.480 | fundamental about once a civilization develops complex enough technology, that technology is

00:26:03.520 | more statistically likely to destroy everybody versus to continue being creative. - That is

00:26:10.560 | absolutely possible. I'm actually putting less credence on that one just because you need it to

00:26:14.960 | happen every single time, right? If even one, I mean, this goes back to von Neumann pointing,

00:26:20.080 | John von Neumann pointed out that you don't need to send the aliens around the galaxy. You can

00:26:26.240 | build self-reproducing probes and send them around the galaxy. And you might think, well,

00:26:31.200 | the galaxy is very big. It's really not. It's some tens of thousands of light years across

00:26:36.960 | and billions of years old. So you don't need to move at a high fraction of the speed of light to

00:26:43.920 | fill the galaxy. - So if you were in an intelligent alien civilization, the dictator of one, you would

00:26:50.000 | just send out a lot of probes, self-replicating probes. - 100%. - Just spread out. - Yes. And

00:26:54.640 | what you should do, so if you want the optimistic spin, here's the optimistic spin. People looking

00:27:00.320 | for intelligent life elsewhere often tune in with their radio telescopes, right? At least we did

00:27:05.920 | before Arecibo was decommissioned. That's not a very promising way to find intelligent life

00:27:14.400 | elsewhere because why in the world would a super intelligent alien civilization waste all of its

00:27:18.640 | energy by beaming it in random directions into the sky? For one thing, it just passes you by,

00:27:26.000 | right? So if we're here on Earth, we've only been listening to radio waves for a couple hundred

00:27:31.200 | years, okay? So if an intelligent alien civilization exists for a billion years,

00:27:37.680 | they have to pinpoint exactly the right time to send us this signal. It is much, much more

00:27:43.200 | efficient to send probes and to park, to go to the other solar systems, just sit there and wait

00:27:52.160 | for an intelligent civilization to arise in that solar system. This is kind of the 2001

00:27:57.520 | monolith hypothesis, right? I would be less surprised to find a sort of quiescent alien

00:28:06.320 | artifact in our solar system than I would to catch a radio signal from an intelligent civilization.

00:28:13.200 | So you're a sucker for in-person conversations versus remote.

00:28:17.440 | I just want to integrate over time. A probe can just sit there and wait,

00:28:23.200 | whereas a radio wave goes right by you.

00:28:25.040 | How hard is it for an alien civilization, again, you're the dictator of one,

00:28:30.320 | to figure out a probe that is most likely to find a common language with whatever it finds?

00:28:38.720 | Couldn't it be like the elected leader of the alien civilization?

00:28:40.400 | Elected leader of a democratic alien civilization, yes.

00:28:46.160 | I think we would figure out that language thing pretty quickly. I mean, maybe not as quickly as

00:28:52.320 | we do when different human tribes find each other because obviously there's a lot of commonalities

00:28:57.280 | in humanity, but there is logic in math and there is the physical world. You can point to a rock and

00:29:03.600 | go rock, right? I don't think it would take that long. I know that Arrival, the movie, based on a

00:29:12.160 | Ted Chiang story, suggested that the way that aliens communicate is going to be fundamentally

00:29:17.520 | different, but also they had precognition and other things I don't believe in. So I think that

00:29:23.040 | if we actually find aliens, that will not be our long-term problem.

00:29:28.160 | So there's a folks, one of the places you're affiliated with is Santa Fe,

00:29:31.440 | and they approach the question of complexity in many different ways and ask the question in many

00:29:36.240 | different ways of what is life, thinking broadly. So do you would be able to find it? You show up,

00:29:44.480 | a probe shows up to a planet, we'll see a thing and be like, "Yeah, that's a living thing."

00:29:50.960 | Well, again, if it's intelligent and technologically advanced, the more short-term

00:29:58.560 | question of if we get some spectroscopic data from an exoplanet, so we know a little bit about

00:30:05.920 | what is in its atmosphere, how can we judge whether or not that atmosphere is giving us

00:30:11.040 | a signature of life existing? That's a very hard question that people are debating about.

00:30:15.520 | I mean, one very simple-minded, but perhaps interesting approach is to say small molecules

00:30:22.640 | don't tell you anything because even if life could make them, something else could also make them.

00:30:27.600 | But long molecules, that's the kind of thing that life would produce.

00:30:31.440 | - So signs of complexity. I don't know. I just have this nervous feeling that we won't be able

00:30:40.000 | to detect. We'll show up to a planet, there'll be a bunch of liquid on it. We take a swim in the

00:30:45.760 | liquid and we won't be able to see the intelligence in it. Whether that intelligence looks like

00:30:54.000 | something like ants or we'll see movement, perhaps strange movement, but we won't be able to

00:31:01.680 | see the intelligence in it or communicate with it. I guess if we have nearly infinite amount of time

00:31:09.120 | to play with different ideas, we might be able to.

00:31:12.480 | - You know, I think, I mean, I'm in favor of this kind of humility, this intellectual humility that

00:31:17.760 | we won't know because we should be prepared for surprises. But I do always keep coming back to

00:31:23.360 | the idea that we all live in the same physical universe. And if, well, let's put it this way,

00:31:31.200 | the development of our intelligence has certainly been connected to our ability to manipulate the

00:31:37.600 | physical world around us. And so I would guess without 100% credence by any means, but my guess

00:31:44.800 | would be that any advanced kind of life would also have that capability. Both dolphins and

00:31:52.960 | octopuses are potential counter examples to that. But I think in the details, there would be enough

00:32:00.480 | similarities that we would recognize it. - I don't know how we got on this topic,

00:32:04.080 | but I think it was from Supermassive Black Holes. So if we return to Black Holes and talk about the

00:32:09.840 | holographic principle more broadly, you have a recent paper on the topic. You've been thinking

00:32:15.200 | about the topic in terms of rigorous research perspective and just as a popular book writer.

00:32:22.800 | So what is the holographic principle? - Well, it goes back to this question

00:32:27.120 | that we were talking about with the information and how it gets out. In quantum mechanics,

00:32:33.840 | certainly, arguably even before quantum mechanics comes along in classical statistical mechanics,

00:32:40.160 | there's a relationship between information and entropy. Entropy is my favorite thing to talk

00:32:45.760 | about, I've written books about and will continue to write books about. So Hawking tells us that

00:32:50.160 | black holes have entropy. And it's a finite amount of entropy, it's not an infinite amount.

00:32:55.760 | But the belief is, and now we're already getting quite speculative, the belief is that the entropy

00:33:02.080 | of a black hole is the largest amount of entropy that you can have in a region of spacetime.

00:33:08.400 | It's sort of the most densely packed that entropy can be. And what that means is there's sort of a

00:33:14.640 | maximum amount of information that you can fit into that region of space and you call it a black

00:33:18.880 | hole. And interestingly, you might expect if I have a box and I'm going to put information in it,

00:33:25.200 | and I don't tell you how I'm going to put the information in, but I ask,

00:33:29.600 | how does the information I can put in scale with the size of the box? You might think, well,

00:33:34.960 | it goes as the volume of the box because the information takes up some volume and I can only

00:33:39.680 | fit in a certain amount. And that is what you might guess for the black hole, but it's not

00:33:43.680 | what the answer is. The answer is that the maximum information as reflected in the black hole entropy

00:33:50.080 | scales as the area of the black hole's event horizon, not the volume inside.

00:33:57.760 | So people thought about that in both deep and superficial ways for a long time and they proposed

00:34:03.440 | what we now call the holographic principle, that the way that spacetime and quantum gravity convey

00:34:10.240 | information or hold information is not different bits or qubits for quantum information at every

00:34:18.480 | point in spacetime. It is something holographic, which means it's sort of embedded in or located

00:34:25.920 | in or can be thought of as pertaining to one dimension less of the three dimensions of space

00:34:32.720 | that we live in. So in the case of the black hole, the event horizon is two-dimensional,

00:34:36.160 | embedded in a three-dimensional universe, and the holographic principle would say all of the

00:34:40.480 | information contained in the black hole can be thought of as living on the event horizon rather

00:34:45.680 | than in the interior of the black hole. I need to say one more thing about that, which is that

00:34:52.000 | this was an idea. The idea I just told you was the original holographic principle

00:34:55.280 | put forward by people like Gerard de Tufte and Leonard Susskind, super famous

00:34:59.920 | physicist. Leonard Susskind was on my podcast and gave a great talk. He's very good at explaining

00:35:06.960 | these things. >> "Mindscape" podcast, everybody should listen.

00:35:09.280 | >> "Mindscape" podcast, that's right, yes. >> And you don't just have physicists on.

00:35:12.640 | >> I don't. >> I love "Mindscape".

00:35:14.960 | >> Oh, thank you very much. >> Curiosity-driven.

00:35:17.280 | >> Yeah, ideas, great ideas from smart people, yeah. But anyway, what I was trying to get at

00:35:21.360 | was Susskind and also de Tufte were a little vague. They were a little hand-wavy about holography and

00:35:26.480 | what it meant. Where holography, the idea that information is sort of encoded on a boundary,

00:35:32.080 | really came into its own was with Juan Maldacena in the 1990s and the ADS-CFD correspondence,

00:35:40.880 | which we don't have to get into that into any detail, but it's a whole full-blown theory.

00:35:46.320 | It's two different theories, one theory in n dimensions of spacetime without gravity,

00:35:52.960 | and another theory in n+1 dimensions of spacetime with gravity. And the idea is that this n

00:35:58.400 | dimensional theory is casting a hologram into the n+1 dimensional universe to make it look like it

00:36:05.440 | has gravity. And that's holography with a vengeance, and that's an enormous source of interest for

00:36:13.440 | theoretical physicists these days. >> How should we picture what impact that has,

00:36:18.960 | the fact that you can store all the information you could think of as all the information that

00:36:24.480 | goes into a black hole can be stored at the event horizon? >> Yeah, I mean, it's a good question.

00:36:28.800 | One of the things that quantum field theory indirectly suggests is that there's not that

00:36:37.840 | much information in you and me compared to the volume of spacetime we take up. As far as quantum

00:36:43.200 | field theory is concerned, you and I are mostly empty space. And so we are not information dense,

00:36:50.640 | right? The density of information in us or in a book or a CD or whatever, computer RAM,

00:36:56.240 | is indeed encoded by volume. Like there's different bits located at different points

00:37:01.680 | in space, but that density of information is super duper low. So we're just like the speed

00:37:07.040 | of light or just like the Big Bang, for the information in a black hole, we are far away

00:37:12.320 | in our everyday experience from the regime where these questions become relevant. So it's very far

00:37:17.520 | away from our intuition. We don't really know how to think about these things. We can do the math,

00:37:21.600 | but we don't feel it in our bones. >> So you can just write off that

00:37:24.880 | weird stuff happens in a black hole. >> Well, we'd like to do better,

00:37:27.920 | but we're trying. I mean, that's why we have an information loss puzzle, because we haven't

00:37:32.800 | completely solved it. So here's just one thing to keep in mind. Once spacetime becomes flexible,

00:37:41.360 | which it does according to general relativity, and you have quantum mechanics, which has

00:37:46.320 | fluctuations and virtual particles and things like that, the very idea of a location in spacetime

00:37:52.320 | becomes a little bit fuzzy, right? Because it's flexible and quantum mechanics says you can't

00:37:56.640 | even pin it down. So information can propagate in ways that you might not have expected. And

00:38:03.840 | that's easy to say and it's true, but we haven't yet come up with the right way to talk about it

00:38:08.640 | that is perfectly rigorous. >> But it's crazy how dense

00:38:11.840 | with information a black hole is. And then plus quantum mechanics starts to come into play. So

00:38:17.520 | you almost want to romanticize the kind of interesting computation type things that are

00:38:22.240 | going on inside the black hole. >> You do, you do. But I'll point

00:38:25.680 | out one other thing. It's information dense, but it's also very, very high entropy.

00:38:32.000 | So a black hole is kind of like a very, very, very specific random number, right? It takes a

00:38:39.440 | lot of digits to specify it, but the digits don't tell you anything. They don't give you anything

00:38:44.400 | useful to work on. So it takes a lot of information, but it's not of a form that we can

00:38:49.920 | learn a lot from. >> But hypothetically,

00:38:53.760 | I guess as you mentioned, the information might be preserved. The information that goes into a

00:38:59.760 | black hole, it doesn't get destroyed. So what does that mean when the entropy is really high?

00:39:04.160 | >> Well, the black hole, I said that the black hole is the highest density of information,

00:39:10.960 | but it's not the highest amount of information because the black hole can evaporate. And when

00:39:15.200 | it evaporates, and people have done the equations for this, when it evaporates, the entropy that

00:39:20.960 | it turns into is actually higher than the entropy of the black hole was, which is good because

00:39:25.280 | entropy is supposed to go up. But it's much more dilute, right? It's spread across a huge volume

00:39:30.880 | of space time. So in principle, all that you made the black hole out of, the information that it

00:39:38.480 | took is still there, we think, in that information, but it's scattered to the four winds.

00:39:43.120 | >> We just talked about the event horizon of a black hole. What's on the inside?

00:39:47.200 | What's at the center of it? >> No one's been there.

00:39:49.280 | >> I came back to tell. >> So again, this is a theoretical

00:39:52.240 | prediction. But I'll say one super crucial feature of the black holes that we know and love,

00:39:58.080 | the kind that Schwarzschild first invented. There's a singularity, but it's not at the middle

00:40:03.120 | of the black hole. Remember, space and time are parts of one unified space time.

00:40:11.280 | The location of the singularity in the black hole is not the middle of space, but our future.

00:40:17.040 | It is a moment of time. It is like a Big Crunch. The Big Bang was an expansion from a singularity

00:40:22.480 | in the past. Big Crunch probably doesn't exist, but if it did, it would be a collapse to a

00:40:27.120 | singularity in the future. That's what the interiors of black holes are like. You can be

00:40:32.480 | fine in the interior, but things are becoming more and more crowded. Space time is becoming

00:40:37.440 | more and more warped, and eventually you hit a limit, and that's the singularity in your future.

00:40:41.760 | >> I wonder what time is like on the inside of a black hole.

00:40:45.200 | >> Time always ticks by at one second per second. That's all it can ever do.

00:40:49.360 | Time can tick by differently for different people. And so you have things like the twin paradox,

00:40:54.640 | where two people initially are the same age. One goes off near the speed of light and comes back.

00:40:59.760 | Now they're not. You can even work out that the one who goes out and comes back will be younger

00:41:04.880 | because they did not take the shortest distance path. But locally, as far as you and your wrist

00:41:11.600 | watch are concerned, time is not funny. Your neurological signals in your brain and your

00:41:19.680 | heartbeat and your wrist watch, whatever's happening to them, is happening to all of

00:41:24.240 | them at the same time. So time always seems to be ticking along at the same rate.

00:41:28.560 | >> Well, if you fall into a black hole, and then I'm an observer just watching it,

00:41:34.960 | and then you come out once it evaporates a million years later,

00:41:40.720 | I guess you'd be exactly the same age. Have you aged at all?

00:41:45.360 | >> You would be converted into photons. You would not be you anymore.

00:41:49.520 | >> Right. So it's not at all possible that information is preserved exactly as it went in.

00:41:54.960 | >> It depends on what you mean by preserved. It's there in the microscopic configuration

00:41:59.600 | of the universe. It's exactly as if I took a regular book, made a paper, and I burned it.

00:42:04.800 | The laws of physics say that all the information in the book is still there in the heat and light

00:42:10.160 | and ashes. You're never going to get it. It's a matter of practice, but in principle, it's still

00:42:15.120 | there. >> But what about the age of things

00:42:17.600 | from the observer perspective, from outside the black hole?

00:42:20.560 | >> From outside the black hole, it doesn't matter because they're inside the black hole.

00:42:26.240 | >> Okay. There's no way to escape the black hole except-

00:42:30.960 | >> To let it evaporate. >> To let it evaporate.

00:42:33.840 | >> But also, by the way, just in relativity, special relativity, forget about general relativity,

00:42:38.960 | it's enormously tempting to say, "Okay, here's what's happening to me right now. I want to know

00:42:46.400 | what's happening far away right now." The whole point of relativity is to say there's no such

00:42:50.800 | thing as right now when you're far away. That is doubly true for what's inside a black hole.

00:42:56.880 | You're tempted to say, "Well, how fast is their clock ticking?" or "How old are they now?"

00:43:01.360 | Not allowed to say that according to relativity. >> Because space and time are treated the same,

00:43:07.280 | and so it doesn't even make sense. What happens to time in the holographic principle?

00:43:11.760 | >> As far as we know, nothing dramatic happens. We're not anywhere close to being confident that

00:43:19.040 | we know what's going on here yet. So there are good unanswered questions about whether time is

00:43:24.000 | fundamental, whether time is emergent, whether it has something to do with quantum entanglement,

00:43:29.840 | whether time really exists at all, different theories, different proponents of different

00:43:36.000 | things. But there's nothing specifically about holography that would make us change our opinions

00:43:41.200 | about time, whatever they happen to be. >> But holography is fundamentally about,

00:43:44.800 | it's a question of space? >> It really is, yeah.

00:43:47.040 | >> Okay, so time is just like a- >> Time just goes along for the ride,

00:43:50.640 | as far as we know, yeah. >> So all the questions about time is just

00:43:52.960 | almost like separate questions, whether it's emergent and all that kind of stuff.

00:43:55.880 | >> Yeah, I mean, that might be a reflection of our ignorance right now, but yes.

00:44:00.000 | >> If we figure out a lot, you know, millions of years from now about black holes,

00:44:04.800 | how surprised would you be if they traveled back in time and told you everything you want to know

00:44:10.240 | about black holes? How much do you think there is still to know? And how mind-blowing would it be?

00:44:17.680 | >> It does depend on what they would say. I think that there are colleagues of mine who think that

00:44:27.200 | we're pretty close to figuring out how information gets out of black holes, how to quantize gravity,

00:44:33.520 | things like that. I'm more skeptical that we are pretty close. I think that there's

00:44:38.000 | room for a bunch of surprises to come. So in that sense, I suspect I would be surprised.

00:44:43.600 | The biggest and most interesting surprise to me would be if quantum mechanics itself

00:44:50.160 | were somehow superseded by something better. As far as I know,

00:44:55.600 | there's no empirical evidence-based reason to think that quantum mechanics is not 100% correct.

00:45:02.480 | But it might not be, that's always possible. And there are, again, respectable friends of mine who

00:45:08.960 | speculate about it. So that's something I would – that's the first thing I would want to know.

00:45:14.560 | >> Oh, so like the black hole would be the most clear illustration.

00:45:18.160 | >> Yeah, that's where it would show up.

00:45:19.360 | >> If there's something, it would show up there.

00:45:21.440 | >> I mean, maybe. The point is that black holes are mysterious for various reasons.

00:45:25.600 | So yeah, if our best theory of the universe is wrong, that might help explain why.

00:45:30.000 | >> Do you think it's possible we'll find something interesting like black holes

00:45:35.120 | sometimes create new universes, or black holes are a kind of portal through space-time to another

00:45:42.160 | place or something like this? And then our whole conception of what is the fabric of space-time

00:45:48.240 | changes completely because black holes, it's like Swiss cheese type of situation.

00:45:51.760 | >> Yeah, you know, that would be less surprising to me because I've already written papers about

00:45:57.360 | that. We don't have, again, strong reason to think that the interior of a black hole leads

00:46:04.720 | to another universe. But it is possible, and it's also very possible that that's true for some black

00:46:09.040 | holes and not others. This is stuff we don't know. It's easy to ask questions we don't know the

00:46:14.160 | answer to. The problem is the questions that are easy to ask that we don't know the answer to are

00:46:18.720 | super hard to answer. >> Because these objects are very

00:46:22.000 | difficult to test and to explore. >> The regimes are just very far away.

00:46:24.960 | So either literally far away in space, but also in energy or mass or time or whatever.

00:46:28.960 | >> You've published a paper on the holographic principle, or that involves the holographic

00:46:34.720 | principle. Can you explain the details of that? >> Yeah, you know, I'm always interested in,

00:46:40.480 | since my first published paper, taking these wild speculative ideas and trying to test them

00:46:46.160 | against data. And the problem is when you're dealing with wild speculative ideas, they're

00:46:50.560 | usually not well-defined enough to make a prediction, right? Like it's kind of a,

00:46:56.800 | I know what's going to happen in some cases, I don't know what's going to happen in other cases.

00:46:59.520 | So we did the following thing. As I've already mentioned, the holographic principle, which is

00:47:06.560 | meant to reflect the information contained in black holes, seems to be telling us that information,

00:47:13.280 | there's less information, less stuff that can go on than you might naively expect.

00:47:19.040 | So let's upgrade naively expect to predict using quantum field theory. Quantum field theory is our

00:47:25.520 | best theory of fundamental physics right now. Unlike this holographic black hole stuff,

00:47:30.400 | quantum field theory is entirely local. In every point of space, something can go on and then you

00:47:36.160 | add up all the different points in space, okay? Not holographic at all. So there's a mismatch

00:47:41.040 | between the expectation for what is happening even in empty space in quantum field theory

00:47:45.840 | versus what the holographic principle would predict. How do you reconcile these two things?

00:47:51.520 | So there's one way of doing it that had been suggested previously, which is to say that

00:47:56.880 | in the quantum field theory way of talking, it implies there's a whole bunch more states,

00:48:03.520 | a whole bunch more ways the system could be than there really are. And just, I'll do a little bit

00:48:10.480 | of math just because there might be some people in the audience who like the math. If I draw

00:48:16.160 | two axes on a two-dimensional geometry, like the surface of the table, right? You know that the

00:48:22.400 | whole point of it being two-dimensional is I can draw two vectors that are perpendicular to each

00:48:26.720 | other. I can't draw three vectors that are all perpendicular to each other, right? They need

00:48:31.520 | to overlap a little bit. That's true for any numbers of dimensions. But I can ask, okay,

00:48:37.600 | how much do they have to overlap? If I try to put more vectors into a vector space

00:48:43.520 | than the dimensionality of the vector space, can I make them almost perpendicular to each other?

00:48:50.080 | And the mathematical answer is, as the number of dimensions gets very, very large,

00:48:55.360 | you can fit a huge extra number of vectors in that are almost perpendicular to each other.

00:49:01.440 | So in this case, what we're suggesting is the number of things that can happen

00:49:06.960 | in a region of space is correctly described by holography. It is somewhat overcounted

00:49:15.120 | by quantum field theory, but that's because the quantum field theory states are not exactly

00:49:20.240 | perpendicular to each other. I should have mentioned that in quantum mechanics, states

00:49:24.560 | are given by vectors in some huge dimensional vector space, very, very, very, very large

00:49:28.320 | dimensional vector space. So maybe the quantum field theory states are not quite perpendicular

00:49:35.040 | to each other. If that is true, that's a speculation already, but if that's true,

00:49:40.560 | how would you know? What is the experimental deviation? And it would have been completely

00:49:46.320 | respectable if we had gone through and made some guesses and found that there is no noticeable

00:49:51.200 | experimental difference because, again, these things are in regimes very, very far away.

00:49:56.480 | We stuck our necks out. We made some very, very specific guesses as to how this weird overlap of

00:50:04.480 | states would show up in the equations of motion for particles like neutrinos. And then we made

00:50:12.960 | predictions on how the neutrinos would behave on the basis of those wild guesses, and then we

00:50:17.680 | compared them with data. And what we found is we're pretty close, but haven't yet reached

00:50:25.280 | the detectability of the effect that we are predicting. In other words, well, basically,

00:50:30.720 | one way of saying what we predict is if a neutrino, and there's reasons why it's neutrinos,

00:50:34.080 | we can go into if you want, but it's not that interesting. If a neutrino comes to us from across

00:50:38.640 | the universe, from some galaxy very, very far away, there is a probability as it's traveling

00:50:44.640 | that it will dissolve into other neutrinos because they're not really perpendicular to each other as

00:50:50.080 | vectors as they would ordinarily be in quantum field theory. And that means that if you look at

00:50:54.320 | neutrinos coming from far enough away with high enough energies, they should disappear. Like if

00:51:01.680 | you see a whole bunch of nearby neutrinos, but then further away, you should see fewer.

00:51:07.520 | And there is an experiment called IceCube, which is this amazing testament to the ingenuity of

00:51:15.440 | human beings where they go to Antarctica and they drill holes and they put photo detectors on a

00:51:22.800 | string a mile deep in these holes. And they basically use all of the ice in a cube, I don't

00:51:30.480 | know whether it's a mile or not, but it's like a kilometer or something like that, some big region,

00:51:34.960 | that much ice is their detector. And they're looking for flashes when a cosmic ray or a

00:51:41.120 | neutrino or whatever hits a ice molecule, water molecule in the ice.

00:51:46.480 | RL: Flashes in the ice, they're looking for.

00:51:48.960 | PW: They're looking for flashes.

00:51:50.000 | RL: But isn't there some crazy, I mean, what does the detector of that look like?

00:51:53.840 | PW: It's a bunch of strings, many, many, many strings with 360-degree photo detectors.

00:52:01.280 | RL: Yeah, that's really cool.

00:52:03.920 | PW: It's extremely cool. And they've done amazing work and they find neutrinos.

00:52:09.200 | RL: So, they're looking for neutrinos.

00:52:10.640 | Yeah. So, the whole point is most cosmic rays are protons. Why? Because protons exist

00:52:17.040 | and they're massive enough that you can accelerate them to very high energies. So,

00:52:22.480 | high-energy cosmic rays tend to be protons. They also tend to hit the Earth's atmosphere

00:52:28.160 | and decay into other particles. So, neutrinos, on the other hand, punch right through,

00:52:33.760 | at least usually, right, to a great extent. So, not just Antarctica, but the whole Earth.

00:52:38.800 | Occasionally, a neutrino will interact with a particle here on Earth, and a neutrino's going

00:52:44.080 | through your body all the time, from the Sun, from the universe, etc. And so, if you're patient

00:52:48.880 | enough and you have a big enough part of the Antarctic ice sheet to look at, the nice thing

00:52:55.040 | about ice is it's transparent. So, you've built yourself, nature has built you a neutrino detector.

00:53:00.400 | PW: So, why ice? So, is it just because of the low noise and you get to watch this thing and it's...

00:53:07.600 | It's much more dense than air, but it's transparent.

00:53:12.160 | LR: So, yeah, much more dense, so higher probability, and then it's transparency,

00:53:16.160 | and then it's also in the middle of nowhere, so you can... Humans are great.

00:53:19.840 | PW: That's all you need. There's not that much ice, right? Yeah. So, there's more ice in Antarctica

00:53:24.160 | than anywhere else, right? So, anyway, you can go and you can get a plot from the IceCube experiment,

00:53:30.320 | how many neutrinos there are that they've detected with very high energies. And we predict in our

00:53:36.800 | weird little holographic guessing game that there should be a cutoff. You should see neutrinos as

00:53:42.000 | you get to higher and higher energies, and then they should disappear. If you look at the data,

00:53:46.000 | their data gives out exactly where our cutoff is. That doesn't mean that our cutoff is right. It

00:53:53.360 | means they lose the ability to do the experiment exactly where we predict the cutoff should be.

00:53:57.920 | LR: Oh, boy. Okay. But why is there a limit?

00:54:02.800 | PW: Oh, just because there are fewer and fewer high-energy neutrinos. So, there's a spectrum,

00:54:08.080 | and it goes down. What we're plotting here is number of neutrinos versus energy. It's fading

00:54:14.080 | away, and they just get very, very few. LR: And you need the high-energy neutrinos

00:54:18.880 | for your prediction. PW: Our effect is a little bit

00:54:21.520 | bigger for higher energies, yeah. And that effect has to do with this almost perpendicular thing.

00:54:26.080 | LR: And let me just mention the name of Oliver Friedrich, who was a postdoc who led this.

00:54:30.080 | He deserves the credit for doing this. I was a co-author and a collaborator. I did some work,

00:54:34.560 | but he really gets the lion's share. PW: Thank you, Oliver. Thank you for

00:54:37.440 | pushing this wild science forward. Just to speak to that, the meta process of it,

00:54:44.240 | how do you approach asking these big questions and trying to formulate it as a paper,

00:54:51.040 | as an experiment that could make a prediction, all that kind of stuff? What's your process?

00:54:55.200 | LR: There's a very interesting thing that happens once you're a theoretical physicist,

00:54:59.680 | once you become trained. You're a graduate student, you've written some papers and whatever.

00:55:03.280 | Suddenly, you are the world's expert in a really infinitesimally tiny

00:55:07.680 | area of knowledge, right? And you know not that much about other areas.

00:55:11.120 | There's an overwhelming temptation to just drill deep, right? Just keep doing basically the thing

00:55:16.880 | that you started doing. But maybe that thing you started doing is not the most interesting thing

00:55:23.040 | to the world or to you or whatever. So you need to separately develop the capability

00:55:28.720 | of stepping back and going, "OK, now that I can write papers in that area,

00:55:33.200 | now that I'm sort of trained enough in the general procedure, what is the best match between my

00:55:40.240 | interests, my abilities, and what is actually interesting?" And honestly, I've not been very

00:55:46.160 | good at that over my career. My process traditionally was I was working in this general

00:55:55.040 | area of particle physics, field theory, general relativity, cosmology. And I would sort of

00:56:02.400 | try to take things other people were talking about and ask myself whether or not it really

00:56:09.120 | fit together. I guess I have three papers that I've ever written that have done super well in

00:56:17.280 | terms of getting cited and things like that. One was my first ever paper that I get very little

00:56:21.760 | credit for. That was my advisor and his collaborator set that up. The other two were

00:56:26.720 | basically my idea. One was right after we discovered that the universe was accelerating.

00:56:32.400 | So in 1998, observations showed that not only is the universe expanding, but it's expanding

00:56:36.720 | faster and faster. So that's attributed to either Einstein's cosmological constant or some more

00:56:43.280 | complicated form of dark energy, some mysterious thing that fills the universe. And people were

00:56:47.920 | throwing around ideas about this dark energy stuff, what could it be and so forth. Most of

00:56:53.120 | the people throwing around these ideas were cosmologists. They work on cosmology. They think

00:56:56.960 | about the universe all at once. Since I like to talk to people in different areas, I was sort of

00:57:04.960 | more familiar than average with what a respectable working particle physicist would think about

00:57:11.520 | these things. And what I immediately thought was, "You guys are throwing around these theories.

00:57:16.720 | These theories are wildly unnatural. They're super finely tuned. Any particle physicist

00:57:20.960 | would just be embarrassed to be talking about this." But rather than just scoffing at them,

00:57:26.880 | I sat down and asked myself, "Okay, is there a respectable version? Is there a way to keep the

00:57:33.360 | particle physicists happy but also make the universe accelerate?" And I realized that there

00:57:38.320 | is some very specific set of models that is relatively natural. And guess what? You can

00:57:44.400 | make a new experimental prediction on the basis of those. And so I did that. People were very happy

00:57:49.680 | about that. >> What was the thing that would make

00:57:51.600 | physicists happy that would make sense of this fragile thing that people call dark energy?

00:57:59.360 | >> So the fact that dark energy pervades the whole universe and is slowly changing,

00:58:06.880 | that should immediately set off alarm bells. Because particle physics is a story of length

00:58:12.640 | scales and time scales that are generally, guess what? Small, right? Particles are small,

00:58:18.400 | they vibrate quickly. And you're telling me now, "I have a new field and its typical rate of change

00:58:24.720 | is once every billion years," right? Like that's just not natural. And indeed, you can formalize

00:58:31.360 | that and say, "Look, even if you wrote down a particle that evolved slowly over billions of

00:58:37.840 | years, if you let it interact with other particles at all, that would make it move faster, its

00:58:45.360 | dynamics would be faster, its mass would be higher," et cetera, et cetera. So there's a whole

00:58:48.560 | story. Things need to be robust and they all talk to each other in quantum field theory. So how do

00:58:53.440 | you stop that from happening? And the answer is symmetry. You can impose a symmetry that protects

00:58:59.440 | your new field from talking to any other fields, okay? And this is good for two reasons. Number

00:59:05.920 | one, it can keep the dynamics slow. So if you just... You can't tell me why it's slow, you just

00:59:11.200 | made that up, but at least it can protect it from speeding up because it's not talking to any other

00:59:15.600 | particles. And the other is it makes it harder to detect. Naively, experiments looking for fifth

00:59:22.240 | forces or time changes of fundamental constants of nature, like the charge of the electron,

00:59:29.520 | these experiments should have been able to detect these dark energy fields. And I was able to

00:59:36.160 | propose a way to stop that from happening. >> The detection.

00:59:39.600 | >> The detection, yeah. Because a symmetry could stop it from interacting with all these other

00:59:44.400 | fields and therefore makes it harder to detect. And just by luck, I realized, because it was

00:59:49.360 | actually based on my first ever paper, there's one loophole. If you impose these symmetries,

00:59:56.320 | so you protect the dark energy field from interacting with any other fields,

00:59:59.600 | there's one interaction that is still allowed that you can't rule out. And it is a very specific

01:00:06.000 | interaction between your dark energy field and photons, which are very common. And it has the

01:00:12.400 | following effect. As a photon travels through the dark energy, the photon has a polarization

01:00:19.120 | up, down, left, right, whatever it happens to be. And as it travels through the dark energy,

01:00:22.880 | that photon will rotate its polarization. This is called birefringence. And you can kind of

01:00:29.680 | run the numbers and say, you know, you can't make a very precise prediction because you're just

01:00:33.040 | making up this model. But if you want to roughly fit the data, you can predict how much polarization

01:00:39.120 | rotation there should be. A couple of degrees, okay? Not that much. So that's very hard to

01:00:45.600 | detect. People have been trying to do it. Right now, literally, we're on the edge of either being

01:00:51.840 | able to detect it or rule it out using the cosmic microwave background. And there is just, you know,

01:00:56.960 | truth in advertising. There is a claim on the market that it's been detected, that it's there.

01:01:03.200 | It's not very statistically significant. If I were to bet, I think it would probably go away. It's

01:01:11.040 | very hard thing to observe. But maybe as you get better and better data, cleaner and cleaner

01:01:16.400 | analysis, it will persist and we will have directly detected the dark energy.

01:01:20.400 | - So if we just take this tangent of dark energy, people will sometimes bring up dark energy and

01:01:28.960 | dark matter as an example why physicists have lost it, lost their mind. We're just going to

01:01:36.720 | say that there's this field that permeates everything. It's unlike any other field and

01:01:41.840 | it's invisible. And it helps us work out some of the math. How do you respond to that?

01:01:48.480 | - Well, two ways. One way is those people would have had to say the same thing when we discovered

01:01:56.080 | the planet Neptune. Because it's exactly analogous where we have a very good theory, in that case,

01:02:03.120 | Newtonian gravity in the solar system. We made predictions. The predictions were slightly off

01:02:08.640 | for the motion of the outer planets. You found that you could explain that motion

01:02:13.760 | by positing something very simple, one more planet in a very, very particular place. And

01:02:19.360 | you went and looked for it and there it was. That was the first successful example of finding dark

01:02:24.240 | matter in the universe. - Matter that we can't see.

01:02:27.120 | - Neptune was dark. - Yeah.

01:02:28.960 | There's a difference between dark matter and dark energy, right? Dark matter, as far as we

01:02:33.200 | are hypothesizing it, is a particle of some sort. It's just a particle that interacts with us very

01:02:39.680 | weakly. So we know how much of it there is. We know more or less where it is. We know some of

01:02:44.240 | its properties. We don't know specifically what it is. But it's not anything fundamentally

01:02:51.600 | mysterious. It's a particle. Dark energy is a different story. So dark energy is indeed

01:02:58.240 | uniformly spread throughout space and has this very weird property that it doesn't seem to evolve

01:03:04.640 | as far as we can tell. It's the same amount of energy in every cubic centimeter of space from

01:03:09.520 | moment to moment in time. That's why far and away the leading candidate for dark energy is

01:03:14.800 | Einstein's cosmological constant. The cosmological constant is strictly constant, 100% constant.

01:03:20.880 | The data say it had better be 98% constant or better. So 100% constant works, right? And it's

01:03:27.680 | also very robust. It's just there. It's not doing anything. It doesn't interact with any other

01:03:31.840 | particles. It makes perfect sense. Probably the dark energy is the cosmological constant.

01:03:35.920 | The dark matter, super important to emphasize here, it was hypothesized at first in the

01:03:43.760 | '70s and '80s mostly to explain the rotation of galaxies. Today, the evidence for dark matter

01:03:53.520 | is both much better than it was in the 1980s and from different sources. It is mostly from

01:04:00.080 | observations of the cosmic background radiation or of large-scale structure. So we have multiple

01:04:06.880 | independent lines of evidence, also gravitational lensing and things like that, many, many pieces

01:04:11.120 | of evidence that say that dark matter is there. And also that say that the effects of dark matter

01:04:17.840 | are different than if we modified gravity. So that was my first answer to your question is

01:04:23.520 | dark matter we have a lot of evidence for. But the other one is, of course, we would love it

01:04:30.960 | if it weren't dark matter. Our vested interest is 100% aligned with it being something more cool and

01:04:38.000 | interesting than dark matter because dark matter is just a particle. That's the most boring thing

01:04:42.240 | in the world. >> And it's non-uniformly

01:04:45.440 | distributed through space, dark matter? >> Absolutely, yeah. You can even see maps of it

01:04:49.120 | that we've constructed from gravitational lensing. >> So verifiable sort of clumps of dark matter in

01:04:54.240 | the galaxy that explains stuff. >> Bigger than the galaxy, sadly. We

01:04:57.920 | think that in the galaxy, dark matter is lumpy, but it's weaker, its effects are weaker. But of

01:05:04.800 | the scale of large-scale structure and clusters of galaxies and things like that, yes, we can

01:05:09.440 | show you where the dark matter is. >> Could there be a super cool

01:05:12.240 | explanation for dark matter that would be interesting as opposed to just another particle

01:05:17.600 | that sits there in clumps? >> The super cool explanation would be

01:05:21.520 | modifying gravity rather than inventing a new particle. Sadly, that doesn't really work. We've

01:05:27.280 | tried, I've tried. That's my third paper that was very successful. I tried to unify dark matter and

01:05:34.320 | dark energy together. That was my idea. That was my aspiration, not even an idea. I tried to do it,

01:05:40.720 | it failed even before we wrote the paper. I realized that my idea did not help. It helps,

01:05:45.760 | it could possibly explain away the dark energy, but it would not explain away the dark matter.

01:05:51.760 | And so I thought it was not that interesting, actually. And then two different collaborators

01:05:56.160 | of mine said, "Has anyone thought of this idea?" They had thought of exactly the same idea

01:05:59.600 | completely independently of me. I said, "Well, if three different people found the same idea,

01:06:04.240 | maybe it is interesting." And so we wrote the paper. And yeah, it was very interesting. People

01:06:08.960 | are very interested in it. >> Can you describe this paper a

01:06:11.440 | little bit? It's just, it's fascinating how much of a thing there is, dark energy and dark matter,

01:06:16.720 | and we don't quite understand it. So what was your dive into exploring how to unify the two?

01:06:22.080 | >> So here is what we know about dark matter and dark energy. They become important in regimes

01:06:30.800 | where gravity is very, very, very weak. That's kind of the opposite from what you would expect

01:06:38.240 | if you actually were modifying gravity. Like there's a rule of thumb in quantum field theory,

01:06:43.360 | et cetera, that new effects show up when the effects are strong, right? We understand weak

01:06:49.360 | fields, we don't understand strong fields. But okay, maybe this is different, right?

01:06:53.440 | So what do I mean by when gravity is weak? The dark energy shows up late in the history of the

01:06:59.120 | universe. Early in the history of the universe, the dark energy is irrelevant. But remember,

01:07:04.320 | the density of dark energy stays constant. The density of matter and radiation go down.

01:07:09.280 | So at early times, the dark energy was completely irrelevant compared to matter and radiation.

01:07:14.560 | At late times, it becomes important. That's also when the universe is dilute and gravity

01:07:19.120 | is relatively weak. Now think about galaxies, okay? A galaxy is more dense in the middle,

01:07:24.720 | less dense on the outside. And there is a phenomenological fact about galaxies that

01:07:28.880 | in the interior of galaxies, you don't need dark matter. That's not so surprising because

01:07:34.000 | the density of stars and gas is very high there, and the dark matter is just subdominant.

01:07:38.160 | But there's generally a radius inside of which you don't need dark matter to fit the data,

01:07:44.960 | outside of which you do need dark matter to fit the data. So that's again when gravity is weak,

01:07:50.080 | right? So I asked myself, of course, we know in field theory, new effects should show up when

01:07:57.920 | fields are strong, not weak, but let's throw that out of the window. Can I write down a theory

01:08:04.320 | where gravity alters when it is weak? And we've already said what gravity is. What is gravity?

01:08:10.800 | It's the curvature of spacetime. So there are mathematical quantities that measure the curvature

01:08:17.280 | of spacetime. And generally, you would say like I have an understanding Einstein's equation,

01:08:22.800 | which I explained to the readers in the book, relates the curvature of spacetime to matter and

01:08:27.920 | energy. The more matter and energy, the more curvature. So I'm saying what if you add a new

01:08:32.720 | term in there that says the less matter and energy, the more curvature? No reason to do that

01:08:41.680 | except to fit the data, right? So I tried to unify the need for dark matter and the need

01:08:47.840 | for dark energy. >> That'd be really cool if that was the case.

01:08:49.920 | >> Super cool, right? It'd be the best. It'd be great. It didn't work.

01:08:53.840 | >> But it'd be really interesting if gravity did something funky when there's not much of it,

01:09:01.600 | almost like at the edges of it, it gets noisy. >> That was exactly the hope, right?

01:09:05.920 | But the great thing about physics is there are equations, right? I mean, you can come up with

01:09:12.560 | the words and you can wave your hands, but then you got to write down the equations, and I did.

01:09:16.720 | And I figured out that it could help with the dark energy, the acceleration of the universe.

01:09:21.840 | It doesn't help with dark matter at all, yeah. >> It just sucks that the scale of galaxies and

01:09:27.040 | scale of solar systems, the physics is kind of boring. >> Yeah, it does. I agree. Again,

01:09:36.320 | that's why it is a little bit, I tear my hair out when people who are not physicists think,

01:09:42.320 | you know, accuse physicists, like you say, of sort of losing the plot because they need dark

01:09:47.760 | matter and dark energy. I don't want dark matter and dark energy. I want something much cooler than

01:09:52.880 | that. I've tried, but you got to listen to the equations and to the data. >> You mentioned three

01:09:59.040 | papers, your first ever, your first awesome paper ever, and your second awesome paper ever. Of

01:10:05.760 | course, you wrote many papers, so you're being very harsh on the others. >> Well, by the way,

01:10:11.200 | this is not awesomeness, this is impact, right? There's no correlation between awesomeness and

01:10:17.680 | impact. Some of my best papers fell without a stone, right? Yeah. The first paper was called

01:10:24.640 | "Limits on the Lorentz and Parity-Violating Modification of Electromagnetism," or

01:10:28.640 | electrodynamics. So, we figured out how to violate Lorentz invariance, which is the symmetry

01:10:33.760 | underlying relativity. And the important thing is we figured out a way to do it that didn't violate

01:10:39.280 | anything else and was experimentally testable, so people love that. The second paper was called

01:10:46.800 | "Quintessence and the Rest of the World." So, quintessence is this dynamical dark energy field.

01:10:52.000 | The rest of the world is because I was talking about how the quintessence field would interact

01:10:56.080 | with other particles and fields and how to avoid the interactions you don't want.

01:11:00.080 | And the third paper was called "Is Cosmic Speedup Due to Gravitational Physics?" Something like that.

01:11:10.560 | So, you see the common theme. I'm taking, you know, what we know, the standard model of particle

01:11:15.040 | physics, general relativity, tweaking them in some way, and then trying to fit the data.

01:11:19.920 | And trying to make it so it's experimentally validated.

01:11:22.560 | Ideally, yes. That's right. That's the goal.

01:11:24.880 | You wrote the book "Something Deeply Hidden on the Mysteries of Quantum Mechanics" and

01:11:29.520 | a new book coming out soon, part of that "Biggest Ideas in the Universe" series we mentioned,

01:11:35.680 | called "Quanta and Fields." So, that's focusing on quantum mechanics.

01:11:41.600 | Big question first. "Biggest Ideas in the Universe."

01:11:44.960 | What to you is most beautiful, or perhaps most mysterious about quantum mechanics?

01:11:51.920 | Quantum mechanics is a harder one. You know, I wrote a textbook on general relativity,

01:11:56.080 | and I started it by saying, "General relativity is the most beautiful physical theory ever

01:12:00.720 | invented." I will stand by that. It is less fundamental than quantum mechanics,

01:12:06.080 | but quantum mechanics is a little more mysterious. So, it's a little bit kludgy right now.

01:12:12.000 | You know, if you think about how we teach quantum mechanics to our students,

01:12:16.560 | the Copenhagen interpretation, it's a god-awful mess. Like, no one's going to accuse that of being

01:12:20.960 | very beautiful. I'm a fan of the many worlds interpretation of quantum mechanics, and that

01:12:26.320 | is very beautiful in the sense that fewer ingredients, just one equation, and it could

01:12:32.160 | cover everything in the world. It depends what you mean by beauty, but I think that the answer

01:12:37.440 | to your question is quantum mechanics can start with extraordinarily austere, tiny ingredients,

01:12:45.280 | and in principle, lead to the world, right? That boggles my mind. It's much more comprehensive.

01:12:53.440 | General relativity is about gravity, and that's great. Quantum mechanics is about everything,

01:12:58.080 | and seems to be up to the task. And so, I don't know, is that beauty or not? But it's

01:13:02.640 | certainly impressive. >> So, both for the theory, the predictive

01:13:04.960 | power of the theory, and the fact that the theory describes tiny things creating everything we see

01:13:09.760 | around us. >> It's a monist theory. In classical mechanics,

01:13:16.320 | I have a particle here, a particle there. I describe them separately. I can tell you what

01:13:20.880 | this particle's doing, what that particle's doing. In quantum mechanics, we have entanglement,

01:13:24.320 | right? As Einstein pointed out to us in 1935. And what that means is there is a single state

01:13:30.640 | for these two particles. There's not one state for this particle, one state for the other particle.

01:13:36.080 | And indeed, there's a single state for the whole universe, called the wave function of the

01:13:40.320 | universe, if you want to call it that. And it obeys one equation, and it is our job then to

01:13:47.600 | sort of chop it up, to carve it up, to figure out how to get tables and chairs and things like that

01:13:52.640 | out of it. >> You mentioned the many worlds

01:13:54.480 | interpretation. And it is, in fact, beautiful. But it's one of your more controversial things,

01:14:02.080 | you stand behind. >> Yeah.

01:14:03.440 | >> You've probably gotten a bunch of flack for it.

01:14:05.200 | >> I'm a big boy, I can take it. >> Well, can you first explain it,

01:14:08.880 | and then maybe speak to the flack you may have gotten?

01:14:11.920 | >> Sure. You know, the classic experiment to explain quantum mechanics to people

01:14:17.600 | is called the Stern-Gerlach experiment. You're measuring the spin of a particle.

01:14:22.640 | Okay? And in quantum mechanics, the spin is, you know, it's just a spin. It's the rate at

01:14:28.720 | which something is rotating around in a very down-to-earth sense, the difference being is

01:14:32.880 | that it's quantized. So for something like a single electron or a single neutron, it's either

01:14:39.120 | spinning clockwise or counterclockwise. Those are the only two, let's put it this way, those are

01:14:43.120 | the only two measurement outcomes you will ever get. There's no, it's spinning faster or slower,

01:14:48.000 | it's either spinning one direction or the other, that's it, two choices, okay?

01:14:52.160 | According to the rules of quantum mechanics, I can set up an electron, let's say, in a state where

01:14:58.480 | it is neither purely clockwise or counterclockwise, but a superposition of both. And that's not just

01:15:06.000 | because we don't know the answer, it's because it truly is both until we measure it. And then

01:15:10.960 | when we measure it, we see one or the other. So this is the fundamental mystery of quantum mechanics

01:15:15.200 | is that how we describe the system when we're not looking at it is different from what we see

01:15:19.360 | when we look at it. So we teach our students in the Copenhagen way of thinking is that the act

01:15:25.280 | of measuring the spin of the electron causes a radical change in the physical state. It spontaneously

01:15:33.520 | collapses from being a superposition of clockwise and counterclockwise to being one or the other.

01:15:40.400 | And you can tell me the probability that that happens, but that's all you can tell me.

01:15:44.800 | And I can't be very specific about when it happens, what caused it to happen,

01:15:48.800 | why it's happening, none of that. That's all called the measurement problem of quantum mechanics.

01:15:53.440 | So many worlds just says, "Look, I just told you a minute ago that there's only one

01:16:00.560 | wave function for the whole universe." And that means that you can't take too seriously

01:16:06.160 | just describing the electron. You have to include everything else in the universe. In particular,

01:16:11.280 | you clearly have to interact with the electron in order to measure it.

01:16:14.720 | So whatever is interacting with the electron should be included in the wave function that

01:16:20.320 | you're describing. And look, maybe it's just you, maybe your eyeballs are able to perceive it,

01:16:24.640 | but okay, I'm going to include you in the wave function. And if you do that, let's be, you know,

01:16:31.280 | since you have a very sophisticated listenership, I'll be a little bit more careful than average.

01:16:35.280 | What does it mean to measure the spin of the electron? We don't need to go into details,

01:16:40.960 | but we want the following thing to be true. If the electron were in a state that was 100%

01:16:47.520 | spinning clockwise, then we want the measurement to tell us it was spinning clockwise. We want

01:16:53.760 | your brain to go, "Yes, the electron was spinning clockwise." Likewise, if it was 100% counter

01:16:59.040 | clockwise, we want to see that, to measure that. The rules of quantum mechanics, the Schrodinger

01:17:05.200 | equation of quantum mechanics is 100% clear that if you want to measure it clockwise when it's

01:17:11.200 | clockwise and measure it counterclockwise when it's counterclockwise, then when it starts out

01:17:16.880 | in a superposition, what will happen is that you and the electron will entangle with each other.

01:17:25.280 | And by that, I mean that the state of the universe evolves into part saying the electron was spinning

01:17:32.160 | clockwise and I saw it clockwise. And part of the state is it's in a superposition with the part

01:17:37.760 | that says the electron was spinning counterclockwise and I saw it counterclockwise.

01:17:41.200 | Everyone agrees with this entirely uncontroversial, straightforward consequence of the Schrodinger

01:17:47.680 | equation. And then Niels Bohr would say, and then part of that wave function disappears.

01:17:53.680 | And we're in the other part and you can't predict which part it will be only the probability.

01:17:58.960 | Hugh Everett, who was a graduate student in the 1950s was thinking about this says,

01:18:02.640 | I have a better idea. Part of the wave function does not magically disappear. It stays there.

01:18:08.320 | The reason why that idea, Everett's idea that the whole wave function always sticks around and just

01:18:14.080 | obeys the Schrodinger equation was not thought of years before is because naively you look at it and

01:18:21.040 | you go, okay, this is predicting that I will be in a superposition, that I will be in a superposition

01:18:28.720 | of having seen the electron be clockwise and having seen it be counterclockwise.

01:18:34.080 | No experimenter has ever felt like they were in a superposition. You always see an outcome.

01:18:38.480 | Everett's move, which was kind of genius was to say, the problem is not the Schrodinger equation.

01:18:46.960 | The problem is you have misidentified yourself in the Schrodinger equation. You have said,

01:18:52.400 | oh, look, there's a person who saw counterclockwise. There's a person who saw

01:18:56.800 | clockwise. I should be that superposition of both. Everett says, no, no, no, you're not.

01:19:03.520 | Because the part of the wave function in which the spin was clockwise, once that exists,

01:19:10.720 | it is completely unaffected by the part of the wave function that says the spin was counterclockwise.

01:19:17.520 | They are apart from each other. They are uninteracting. They have no influence. What

01:19:23.040 | happens in one part has no influence in the other part. So Everett says the simple resolution is to

01:19:28.160 | identify yourself as either the one who saw spin clockwise or the one who saw spin counterclockwise.

01:19:36.320 | There are now two people. Once you've done that experiment, the Schrodinger equation doesn't have

01:19:41.440 | to be messed with. All you have to do is locate yourself correctly in the wave function. That's

01:19:46.320 | many worlds. >> The number of worlds is-

01:19:49.760 | >> Very big. >> Very, very, very big.

01:19:53.280 | Where do those worlds fit? Where do they go? >> So the short answer is the worlds don't exist

01:20:03.280 | in space. Space exists separately in each world. So I mean, there's a technical answer to your

01:20:12.080 | question, which is Hilbert space, the space of all possible quantum mechanical states.

01:20:16.320 | But physically, we want to put these worlds somewhere. That's just a wrong

01:20:21.120 | intuition that we have. There is no such thing as the physical spatial location of the worlds

01:20:27.360 | because space is inside the worlds. >> One of the properties of this

01:20:31.120 | interpretation is that you can't travel from one world to the other.

01:20:34.400 | >> That's right. >> Which kind of makes you feel

01:20:37.440 | that they're existing separately. >> They are existing separately

01:20:44.320 | and simultaneously. >> And simultaneously.

01:20:46.720 | >> Without locations in space. >> Without locations in space. How is it

01:20:51.200 | possible to visualize them existing without a location in space?

01:20:54.720 | >> The real answer to that, the honest answer is the equations predict it.

01:20:59.920 | >> Yeah, yeah. >> If you can't visualize it,

01:21:03.200 | so much worse for you. The equations are crystal clear about what they're predicting.

01:21:07.280 | >> Is there a way to get to the closer to understanding and visualizing the weirdness

01:21:13.360 | of the implications of this? >> You know, I don't think it's that

01:21:16.720 | hard. It wasn't that hard for me. I don't mind the idea that when I make a quantum mechanical

01:21:24.240 | measurement, there is, later on in the universe, multiple descendants of my present self who got

01:21:30.560 | different answers for that measurement. I can't interact with them. Hilbert space,

01:21:36.880 | the space of all quantum wave functions, was always big enough to include all of them.

01:21:42.400 | I'm going to worry about the parts of the universe I can observe. Let's put it this way.

01:21:48.160 | Many worlds comes about by taking the Schrodinger equation seriously. The Schrodinger equation was

01:21:54.640 | invented to fit the data, to fit the spectrum of different atoms and different emission and

01:22:00.160 | absorption experiments. It's perfectly legitimate to say, "Well, okay, you're taking the Schrodinger

01:22:08.000 | equation. You're extrapolating it. You're trusting it, believing it beyond what we can observe.

01:22:15.120 | I don't want to do that." That's perfectly legit, except, okay, then what do you believe?

01:22:21.360 | Come up with a better theory. You're saying you don't believe the Schrodinger equation.

01:22:27.200 | Tell me the equation that you believe in. Turns out, and people have done that,

01:22:31.360 | turns out it's super hard to do that in a legitimate way that fits the data.

01:22:36.640 | >> And many worlds is a really clean-

01:22:38.960 | >> Absolutely. The most austere, clean, no extra baggage theory of quantum mechanics.

01:22:44.800 | >> But if it, in fact, is correct, isn't this the weirdest thing of anything we know?

01:22:54.480 | >> Yes. In fact, let me put it this way. The single best reason in my mind to be skeptical

01:23:02.720 | about many worlds is not because it doesn't make sense, or it doesn't fit the data,

01:23:07.840 | or I don't know where the worlds are going, or whatever. It's because to make that extrapolation,

01:23:14.640 | to take seriously the equation that we know is correct in other regimes requires new philosophy,

01:23:20.400 | requires a new way of thinking about identity, about probability, about prediction, a whole

01:23:26.960 | bunch of things. It's worked to do that philosophy, and I've been doing it, and others have done it,

01:23:32.080 | and I think it's very, very doable, but it's not straightforward. It's not a simple

01:23:38.720 | extrapolation from what we already know. It's a grand extrapolation very far away.

01:23:43.760 | And if you just wanted to be methodologically conservative and say, "That's a step too far,

01:23:50.880 | I don't want to buy it," I'm sympathetic to that. I think that you're just wimping out.

01:23:55.280 | I think that you should have more courage, but I get the impulse.

01:23:59.920 | >> And there is, under many worlds, an arrow of time where if you rewind it back,

01:24:08.480 | there's going to be one initial state. >> That's right. All of quantum mechanics,

01:24:14.800 | all different versions require a kind of arrow of time. It might be different in every kind.

01:24:18.960 | But the quantum measurement process is irreversible. You can measure something,

01:24:26.160 | it collapses, you can't go backwards. If someone tells you the outcome, if I say,

01:24:29.280 | "I've measured an electron, its spin is clockwise," and they say, "What was it before I measured it?"

01:24:35.040 | You know there was some part of it that was clockwise, but you don't know how much, right?

01:24:39.520 | In many worlds, it's no different. But the nice thing is that the kind of arrow of time you need

01:24:45.200 | in many worlds is exactly the kind of arrow of time you need anyway for entropy and thermodynamics

01:24:50.960 | and so forth. You need a simple, low-entropy initial state. That's what you need in both cases.

01:24:55.680 | >> So if you actually look at under many worlds into the entire history of the universe,

01:25:01.120 | correct me if I'm wrong, but it looks very deterministic.

01:25:05.440 | >> Yes.

01:25:06.240 | >> In each moment, does the moment contain the memory of the entire history of the universe?

01:25:11.440 | To you, does the moment contain the memory of everything that preceded it?

01:25:16.480 | >> As far as we know, so according to many worlds, the wave function of the universe,

01:25:23.120 | all the branches of the universe at once, all the worlds, does contain all the information.

01:25:27.040 | Calling it a memory is a little bit dangerous because it's not the same kind of memory that

01:25:35.360 | you and I have in our brains because our memories rely on the arrow of time.

01:25:39.920 | And the whole point of the Schrodinger equation or Newton's laws is they don't have an arrow of time

01:25:47.520 | built in. They're reversible. The state of the universe not only remembers where it came from,

01:25:53.280 | but also determines where it's going to go in a way that our memories don't do that.

01:25:56.880 | >> But our memories, we could do replay. Can you do this?

01:26:00.720 | >> We can, but the act of forming a memory increases the entropy of the universe.

01:26:04.960 | It is an irreversible process also, right? You can walk on a beach and leave your footprints

01:26:11.440 | there. That's a record of your passing. It will eventually be erased by the ever-increasing

01:26:16.640 | entropy of the universe. >> But you can imperfectly replay it.

01:26:20.000 | >> Yes. >> I guess, can we return,

01:26:22.960 | travel back in time imperfectly? >> Oh, it depends on the level of

01:26:28.960 | precision you're trying to ask that question. The universe contains the information about where the

01:26:35.520 | universe was, but you and I don't. We're nowhere close.

01:26:38.480 | >> And it's, what, computationally very costly to try to consult the universe?

01:26:44.880 | >> Well, it depends on, again, exactly what you're asking. There are some simple questions,

01:26:49.840 | like what was the temperature of the universe 30 seconds after the Big Bang? We can answer that,

01:26:56.320 | right? That's kind of amazing that we can answer that to pretty high precision,

01:27:01.440 | but if you want to know where every atom was, then no.

01:27:04.720 | >> What to you is the Big Bang? Why did it happen?

01:27:12.400 | >> We have no idea. I think that that's a super important question that I can imagine making

01:27:18.560 | progress on, but right now I'm more or less maximally uncertain about what the answer is.

01:27:23.920 | >> Do you think black holes will help, potentially? >> No, not that much. Quantum gravity will help,

01:27:30.000 | and maybe black holes will help us figure out quantum gravity, so indirectly, yes. But we have

01:27:35.360 | a situation where general relativity, Einstein's theory, unambiguously predicts there was a

01:27:41.040 | singularity in the past. There was a moment of time when the universe had infinite curvature,

01:27:47.200 | infinite energy, infinite expansion rate, the whole bit. That's just a fancy way of saying

01:27:53.280 | the theory has broken down, and classical general relativity is not up to the task of saying what

01:27:58.240 | really happened at that moment. So it is completely possible there was, in some sense,

01:28:03.760 | a moment of time before which there were no other moments, and that would be the Big Bang. Even if

01:28:09.680 | it's not a classical general relativity kind of thing, even if quantum mechanics is involved,

01:28:13.440 | maybe that's what happened. It's also completely possible there was time before that,

01:28:17.840 | space and time, and they evolved into our hot Big Bang by some procedure that we don't really

01:28:23.440 | understand. >> And if time and space are emergent,

01:28:25.920 | then the before even starts getting real weird. >> Well, I think that if there is a first moment

01:28:33.120 | of time, that would be very good evidence, or that would fit hand in glove with the idea that

01:28:38.400 | time is emergent. If time is fundamental, then it tends to go forever, because it's fundamental.

01:28:43.600 | >> Well, yeah. I mean, the general formulation of this question is what's outside of it,

01:28:48.800 | what's outside of our universe? So in time and in space. I know it's a pothead question, Sean.

01:28:54.960 | I understand. I apologize. >> Look, that's my life. My life

01:28:59.040 | is asking pothead questions. Some of them, the answer is, that's not the right way to think

01:29:02.800 | about it. >> Okay, but is it possible to think at all

01:29:05.920 | about what's outside our universe? >> It's absolutely legit to ask questions,

01:29:12.240 | but you have to be comfortable with the possibility that the answer is, there's no such thing as

01:29:16.160 | outside our universe. That's absolutely on the table. In fact, that is the simplest,

01:29:20.160 | most likely to be correct answer that we know of. >> But it's the only thing

01:29:24.960 | in the universe that wouldn't have an outside. >> Yeah, if the universe is the totality of

01:29:32.720 | everything, it would not have an outside. >> That's so weird to think that there's

01:29:36.240 | not an outside. We want there to be. We want there to be sort of a creator, a creative force

01:29:45.120 | that led to this. An outside, like this is our town, and then there's a bigger world,

01:29:50.160 | and there's always a bigger world. >> Because that is our experience. That's

01:29:54.080 | the world we grew up in, right? The universe doesn't need to obey those rules.

01:29:59.840 | >> Such a weird thing. >> When I was a kid,

01:30:02.960 | that used to keep me up at night. Like, what if the universe had not existed?

01:30:05.840 | >> Right. And it feels like a lot of pressure that if this is the only universe,

01:30:13.200 | and we're here, one of the few intelligent civilizations, maybe the only one,

01:30:19.920 | it's the old theories that we're the center of everything. It just feels suspicious.

01:30:25.040 | That's why many worlds is kind of exciting to me, because it's humbling in all the right

01:30:30.560 | kinds of ways. It feels like infinity is the way this whole thing runs.

01:30:36.000 | >> There's one pitfall that I'll just mention, because there's a move that is made in these

01:30:42.880 | theoretical edges of cosmology that I think is a little bit mistaken, which is to say,

01:30:46.480 | I'm going to think about the universe on the basis of imagining that I am a typical observer.

01:30:54.720 | I, this is called the principle of typicality, or the principle of mediocrity, or even the

01:30:58.640 | Copernican principle. Nothing special about me, I'm just typical in the universe. But then you

01:31:04.240 | draw some conclusions from this, and what you end up realizing is you've been hilariously presumptuous,

01:31:11.040 | because by saying I'm a typical observer in the universe, you're saying typical observers in the

01:31:15.360 | universe are like me. And that is completely unjustified by anything. So, I'm not telling

01:31:21.920 | you what the right way to do it is, but these kinds of questions that are not quite grounded in

01:31:26.480 | experimental verification or falsification are ones you have to be very careful about.

01:31:33.040 | >> That to me is one of the most interesting questions. And there's different ways to

01:31:38.400 | approach it, but like, what's outside of this? How did the big mess start? How do we get something

01:31:43.440 | from nothing? That's always the thing you're sneaking up to. When you're studying all of

01:31:50.720 | these questions, you're always sneaking, that's where the black hole and the unifying, getting

01:31:54.640 | quantum gravity, all this kind of stuff, you're always sneaking up to that question, where did

01:32:00.000 | all of this come from? And I think that's probably an answerable question, right?

01:32:07.040 | >> No.

01:32:09.200 | >> It doesn't have to be. So, you think there could be a turtle at the bottom of this

01:32:13.840 | that refuses to reveal its identity?

01:32:16.960 | >> Yes. I think that specifically the question, why is there something rather than nothing,

01:32:23.040 | does not have the kind of answer that we would ordinarily attribute to why questions.

01:32:29.360 | Because typical why questions are embedded in the universe, and when we answer them,

01:32:37.600 | we take advantage of the features of the universe that we know and love.

01:32:40.560 | But the universe itself, as far as we know, is not embedded in anything

01:32:44.400 | bigger or stronger, and therefore, it can just be.

01:32:46.960 | >> Do you think it's possible this whole place is simulated?

01:32:50.720 | >> Sure.

01:32:51.760 | >> It's a really interesting, dark, twisted video game that we're all existing in.

01:32:56.720 | >> You know, my own podcast listeners, Mindscape listeners, tease me because they know from my

01:33:02.880 | AMA episodes that if you ever start a question by asking, "Do you think it's possible that?"

01:33:10.080 | The answer is going to be yes. That might not be the answer that you care about, but it's possible,

01:33:16.560 | sure, as long as you're not, you know, adding two even numbers together and getting an odd number.

01:33:21.200 | >> When you say it's possible, there's a mathematically yes,

01:33:24.560 | and then there's more of like intuitive...

01:33:26.320 | >> Yeah. You want to know whether it's plausible. You want to know is there a

01:33:30.240 | reasonable non-zero credence to attach to this. I don't think that there's any

01:33:38.000 | philosophical knockout objection to the simulation hypothesis. I also think that

01:33:42.640 | there's absolutely no reason to take it seriously. >> Do you think humans will try to create one?

01:33:47.680 | I guess that's how I always think about it. You know, I've spent quite a bit of time

01:33:53.680 | over the past few years and a lot more recently in virtual worlds, and just am always captivated

01:34:03.200 | by the possibility of creating higher and higher resolution worlds, and as we'll talk a little bit

01:34:08.480 | about artificial intelligence, sort of the advancement on the Sora front, that you can

01:34:14.800 | automatically generate those worlds, and the possibility of existing in those automatically

01:34:20.400 | generated worlds is pretty exciting, as long as there's a consistent physics, quantum mechanics,

01:34:25.680 | and general relativity that governs the generation of those worlds. So it just seems like humans will

01:34:32.640 | for sure try to create this. >> Yeah, I think they will create

01:34:36.400 | better and better simulations. I think the philosopher David Chalmers has done what I

01:34:40.800 | consider to be a good job of arguing that we should treat things that happen in virtual reality

01:34:45.440 | and in simulated realities as just as real as the reality that we experience. I also think that as a

01:34:51.760 | practical matter, people will realize how much harder it is to simulate a realistic world than we

01:34:58.000 | naively believe, so this is not a my lifetime kind of worry.

01:35:01.440 | >> Yeah, the practical matter of going from a sort of a prototype that's impressive

01:35:06.480 | to a thing that governs everything. Similar question on this front is in AGI. Yeah,

01:35:14.320 | you said that we're very far away from AGI. >> I want to eliminate the phrase AGI.

01:35:20.080 | >> So basically, when you're analyzing large language models and seeing how far are they

01:35:26.960 | from whatever AGI is, and we could talk about different notions of intelligence that we're not

01:35:32.480 | as close as kind of some people in public view are talking about. So what's your intuition behind

01:35:40.400 | that? >> My intuition is basically that

01:35:42.880 | artificial intelligence is different than human intelligence, and so the mistake that is being

01:35:50.000 | made by focusing on AGI, among those who do, is an artificial agent, as we can make them now or in

01:35:57.840 | the near future, might be way better than human beings at some things, way worse than human

01:36:04.400 | beings at other things. And rather than trying to ask how close is it to being a human-like

01:36:09.680 | intelligence, we should appreciate it for what its capabilities are, and that will both be more

01:36:15.040 | accurate and help us put it to work and protect us from the dangers better, rather than always

01:36:20.560 | anthropomorphizing it. >> I think the underlying idea there,

01:36:25.440 | under the definition of AGI, is that the capabilities are extremely impressive.

01:36:33.600 | That's not a precise statement, but meaning— >> No, I get that. I completely agree.

01:36:37.760 | >> And then the underlying question, where a lot of the debate is, is how impressive is it?

01:36:43.680 | What are the limits of large-language models? Can they really do things like common-sense

01:36:48.800 | reasoning? How much do they really understand about the world? Are they just fancy mimicry

01:36:53.840 | machines? And where do you fall on that, as to the limits of large-language models?

01:37:02.400 | >> I don't think that there are many limits in principle. I'm a physicalist about consciousness

01:37:10.320 | and awareness and things like that. I see no obstacle to, in principle, building an artificial

01:37:15.760 | machine that is indistinguishable in thought and cognition from a human being. But we're not trying

01:37:21.360 | to do that, right? What a large-language model is trying to do is to predict text. That's what it

01:37:28.000 | does. And it is leveraging the fact that we human beings, for very good evolutionary biology reasons,

01:37:37.520 | attribute intentionality and intelligence and agency to things that act like human beings.

01:37:43.360 | As I was driving here to get to this podcast space, I was using Google Maps, and Google Maps

01:37:51.440 | was talking to me. But I wanted to stop to get a cup of coffee, so I didn't do what Google Maps

01:37:57.280 | told me to do. I went around a block that it didn't like. And so it gets annoyed, right? It

01:38:04.080 | says, "No, why are you doing?" It doesn't say exactly in this, but you know what I mean. It's

01:38:07.760 | like, "No, turn left, turn left," and you turn right. It is impossible as a human being not to

01:38:13.680 | feel a little bit sad that Google Maps is getting mad at you. It's not. It's not even trying to.

01:38:19.120 | It's not a large-language model. There's no aspirations to intentionality, but we attribute

01:38:24.400 | that all the time. Dan Dennett, the philosopher, wrote a very influential paper on the intentional

01:38:31.440 | stance, the fact that it's the most natural thing in the world for we human beings to attribute more

01:38:38.000 | intentionality to artificial things than are really there, which is not to say it can't be really

01:38:43.680 | there. But if you're trying to be rational and clear-thinking about this, the first step is to

01:38:49.760 | recognize our huge bias towards attributing things below the surface to systems that are able to,

01:38:59.200 | at the surface level, act human. - So if that huge bias of intentionality is there in the data,

01:39:05.600 | in the human data, in the vast landscape of human data that AI models, large-language models, and

01:39:12.160 | video models in the future are trained on, don't you think that that intentionality will emerge

01:39:20.080 | as fundamental to the behavior of these systems naturally? - Well, I don't think it will happen

01:39:25.760 | naturally. I think it could happen. Again, I'm not against the principle. But again, the way that

01:39:33.120 | large-language models came to be and what they're optimized for is wildly different than the way

01:39:40.400 | that human beings came to be and what they're optimized for. So I think we're missing a chance

01:39:48.400 | to be much more clear-headed about what large-language models are by judging them against

01:39:54.000 | human beings, again, both in positive ways and negative ways. - Well, I think sort of to push

01:39:58.240 | back on what they're optimized for, it's different to describe how they're trained versus what they're

01:40:02.240 | optimized for. So they're trained in this very trivial way of predicting text tokens. But you

01:40:09.040 | can describe what they're optimized for and what the actual task at hand is, is to construct a

01:40:13.760 | world model, meaning an understanding of the world. And that's where it starts getting closer to what

01:40:19.920 | humans are kind of doing. We're just, in the case of large-language models, know how the sausage is

01:40:25.360 | made and we don't know how it's made for us humans. - But they're not optimized for that.

01:40:29.520 | They're optimized to sound human. - That's the fine-tuning. But the actual training is optimized

01:40:35.600 | for understanding, creating a compressed representation of all the stuff that humans

01:40:43.120 | have created on the internet. And the hope is that that gives you a deeper understanding

01:40:49.120 | of the world. - Yeah. So that's why I think that there's a set of hugely interesting questions to

01:40:54.080 | be asked about the ways in which large-language models actually do represent the world. Because

01:41:01.200 | what is clear is that they're very good at acting human. The open question in my mind is,

01:41:08.560 | is the easiest, most efficient, best way to act human to do the same things that human beings do?

01:41:15.120 | Or are there other ways? And I think that's an open question. I just heard a talk by Melanie

01:41:20.080 | Mitchell at Santa Fe Institute, an artificial intelligence researcher. And she told two stories

01:41:25.760 | about two different papers, one that someone else wrote and one that her group is following up on.

01:41:29.680 | And they were modeling Othello. Othello, the game with a little rectangular board,

01:41:34.240 | white and black squares. So the experiment was the following. They fed a neural network

01:41:39.840 | the moves that were being made in the most symbolic form, like E5. Just means that,

01:41:46.800 | okay, you put a token down E5. So it gives a long string. It does this for millions of games,

01:41:51.440 | right? Real legitimate games. And then it asks the question, the paper asks the question, okay,

01:41:56.320 | you've trained it to tell what would be a legitimate next move from not a legitimate next

01:42:02.720 | move. Did it in its brain, in its little large language model brain, I don't even know if it's

01:42:09.040 | technically a large language model, but a deep learning network. Did it come up with a representation

01:42:13.920 | of the Othello board? Well, how do you know? And so they construct a little probe network that they

01:42:19.600 | insert and you ask it, what is it doing right at this moment, right? And the answer is that the

01:42:26.320 | little probe network can ask, would this be legitimate or is this token white or black or

01:42:31.040 | whatever? Things that in practice would amount to it's invented the Othello board. And it found that

01:42:40.800 | the probe got the right answer, not 100% of the time, but more than by chance, substantially more

01:42:47.360 | than by chance. So they said, there's some tentative evidence that this neural network

01:42:54.240 | has discovered the Othello board just out of data, raw data, right? But then Melanie's group

01:43:00.000 | asked the question, okay, are you sure that that understanding of the Othello board wasn't built

01:43:07.120 | into your probe? And what they found was like at least half of the improvement was built into the

01:43:13.040 | probe, not all of it, right? And look, a Othello board is way simpler than the world. So that's

01:43:23.360 | why I just think it's an open question whether or not the, I mean, it would be remarkable either way

01:43:31.120 | to learn that large language models that are good at doing what we train them to do are good because

01:43:37.840 | they've built the same kind of model of the world that we have in our minds or that they're good,

01:43:42.960 | despite not having that model. Either one of these is an amazing thing. I just don't think

01:43:46.720 | the data are clear on which one is true. - I think I have some sort of intellectual

01:43:51.520 | humility about the whole thing because I was humbled by several stages in the machine learning

01:43:56.880 | development over the past 20 years. And I just would never have predicted that LLMs, the way

01:44:06.720 | they're trained, on the scale of data they're trained would be as impressive as they are.

01:44:11.040 | And there, that's where intellectual humility steps in, where my intuition would say something

01:44:17.120 | like with Melanie, where you need to be able to have very sort of concrete, common sense reasoning,

01:44:22.720 | symbolic reasoning type things in a system in order for it to be very intelligent. But here,

01:44:30.640 | I'm so impressed by what it's capable to do, train on the next token prediction, essentially,

01:44:36.960 | that I just, my conception of the nature of intelligence is just completely, not completely,

01:44:45.920 | but humbled, I should say. - Look, and I think that's perfectly fair. I also

01:44:50.480 | was, I almost say pleasantly, I don't know whether it's pleasantly or unpleasantly,

01:44:55.440 | but factually surprised by the recent rate of progress. Clearly, some kind of phase transition

01:45:00.640 | percolation has happened, right? And the improvement has been remarkable, absolutely

01:45:05.040 | amazing. That I have no arguments with. That doesn't yet tell me the mechanism by which that

01:45:14.480 | improvement happened. Constructing a model much like a human being would have is clearly one

01:45:19.840 | possible mechanism, but part of the intellectual humility is to say maybe there are others.

01:45:24.000 | - I was chatting with the CEO of Anthropic, Dario Medet, so behind Klon, and that company,

01:45:32.320 | but a lot of the AI companies are really focused on expanding the scale of compute.

01:45:38.000 | Sort of, if we assume that AI is not data limited, but is compute limited,

01:45:44.480 | you can make the system much more intelligent by using more compute.

01:45:49.680 | So let me ask you, almost on the physics level, do you think physics can help

01:45:57.360 | expand the scale of compute, and maybe the scale of energy required to make that compute happen?

01:46:02.240 | - Yeah, 100%. I think this is like one of the biggest things that physics can help with,

01:46:08.080 | and it's an obvious kind of low-hanging fruit situation where the heat generation,

01:46:15.760 | the inefficiency, the waste of existing high-level computers is nowhere near the efficiency of our

01:46:25.040 | brains. It's hilariously worse. And we kind of haven't tried to optimize that hard on that

01:46:30.160 | frontier. I mean, your laptop heats up when you're sitting on your lap, right? It doesn't need to,

01:46:34.240 | your brain doesn't heat up like that. So clearly there exists in the world of physics the capability

01:46:41.520 | of doing these computations with much less waste heat being generated,

01:46:45.760 | and I look forward to people doing that, yeah. - Are you excited for the possibility of nuclear

01:46:50.480 | fusion? - I am cautiously optimistic,

01:46:53.680 | excited it'll be too strong. I mean, it'd be great, right? But if we really tried solar power,

01:46:59.600 | it would also be great. - I think Elias Discover said this,

01:47:03.840 | that the future of humanity on Earth would be just the entire surface of Earth is covered in

01:47:10.560 | solar panels and data centers. - Why would you waste the surface

01:47:13.760 | of the Earth with solar panels? Put them in space. - Sure, you can go in space, yeah.

01:47:17.520 | - Space is bigger than the Earth. - Yeah, just solar panels everywhere.

01:47:21.440 | - Yeah. - I like it.

01:47:24.080 | - We already have fusion. It's called the sun. - Yeah, that's true. And there's probably

01:47:29.520 | more and more efficient ways of catching that energy.

01:47:32.800 | - Sending it down is the hard part, absolutely. But that's an engineering problem, yeah.

01:47:37.440 | - So I just wonder where data centers, the compute centers can expand to. If that's the future,

01:47:44.480 | if AI is as effective as it possibly could be, then the scale of computation will keep increasing.

01:47:52.480 | Perhaps it's a race between efficiency and scale. - There are constraints, right? There's a certain

01:47:58.800 | amount of energy, a certain amount of damage we can do to the environment before it is not

01:48:02.320 | worth it anymore. So yeah, I think that's a new question. In fact, it's kind of frustrating

01:48:07.120 | because we get better and better at doing things efficiently, but we invent more things we want to

01:48:12.640 | do faster than we get good at doing them efficiently. So we're continuing to make

01:48:18.000 | things worse in various ways. - I mean, that's the dance of humanity,

01:48:21.600 | where we're constantly creating better and better and better technologies that are potentially

01:48:25.760 | causing a lot more harm. And that includes for weapons, it includes AI used as weapons,

01:48:30.960 | that includes nuclear weapons, of course. Which is surprising to me that we haven't destroyed

01:48:36.240 | human civilization yet, given how many nuclear warheads are out there.

01:48:40.720 | - Look, I'm with you. Between nuclear and bioweapons, it is a little bit surprising

01:48:47.040 | that we haven't caused enormous devastation. Of course, we did drop two atomic bombs on Japan,

01:48:51.440 | but compared to what could have happened, or could happen tomorrow, it could be much worse.

01:48:56.800 | - Yeah, it does seem like there's an underlying, speaking of quantum fields,

01:49:01.760 | there's like a field of goodness within the human heart that, in some kind of game-theoretic way,

01:49:11.120 | would create really powerful things that could destroy each other. And there's greed and ego

01:49:15.040 | and all this kind of power-hungry dictators that are at play here in all the geopolitical landscape,

01:49:22.240 | but we somehow always don't go too far. - Yeah, but that's exactly what you would say

01:49:26.880 | right before we went too far. - Right before we went too far.

01:49:29.120 | And that's why we don't see aliens. So you're, like I mentioned, associated with Santa Fe

01:49:36.000 | Institute. I just would love to take a stroll down the landscape of ideas explored there.

01:49:43.200 | - Sure. - So they look at complexity

01:49:45.280 | in all kinds of ways. What do you think about the emergence of complexity from simple things

01:49:50.800 | interacting simply? - I think it's a fascinating topic. I mean,

01:49:54.240 | that's why I'm thinking about these things these days rather than the papers that I was describing

01:49:58.320 | to you before. All of those papers I described to you before are guesses. Like, what if the laws of

01:50:04.720 | physics are different in the following way? And then you can work out the consequences. At some

01:50:08.400 | point in my life, I said, like, what is the chance I'm gonna guess right? Einstein guessed right,

01:50:12.160 | Steven Weinberg guessed right, but there's a very small number of times that people guessed right.

01:50:16.320 | Whereas with this emergence of complexity from simplicity, I really do think that we haven't

01:50:22.960 | understood the basics yet. I think we're still kind of pre-paradigmatic. There've been some

01:50:28.240 | spectacular discoveries. People like Jeffrey West at Santa Fe and others have really given us true

01:50:35.760 | insights into important systems, but still there's a lot of the basics I think are not understood.

01:50:40.800 | And so searching for the general principles is what I like to do. And I think it's absolutely

01:50:46.400 | possible that, I mean, to be a little bit more substantive than that, I think this is kind of

01:50:51.600 | a cliche, I think the key is information. And I think that what we see through the history of the

01:50:57.920 | universe as you go from simple to more and more complex is really subsystems of the universe

01:51:04.560 | figuring out how to use information to do whatever, to survive or to thrive or to reproduce.

01:51:10.320 | I mean, that's the sort of fuel, the leverage, the resource that we have for a while anyway,

01:51:16.400 | until the heat death, but that's where the complexity is really driven by.

01:51:20.240 | - Yeah, but the mechanism of it, I mean, you mentioned Jeffrey West, what are interesting

01:51:24.560 | inklings of progress in this realm and what are systems that interest you in terms of information?

01:51:30.160 | So, I mean, for me just as a fan of complexity, just even looking at simple cellular automata

01:51:36.560 | is always just a fascinating way to illustrate the emergence of complexity.

01:51:42.240 | - So for those of the listeners who don't know, viewers, cellular automata come from

01:51:48.080 | imagining a very simple configuration, for example, a set of ones and zeros along a line.

01:51:55.040 | And then you met a rule that says, okay, I'm gonna evolve this in time. And generally the

01:52:00.400 | simplest ones start with just each block of three ones and zeros have a rule that they will

01:52:06.720 | determinously go to either a one or a zero. And you can actually classify all the different

01:52:10.960 | possibilities, a small number of possible cellular automata of that form. And what was discovered

01:52:16.480 | by various people, including Stephen Wolfram is some of these cellular automata have the feature

01:52:22.240 | that you start from almost nothing like zero, zero, zero, zero, one, zero, zero, zero, zero,

01:52:27.120 | and you let it rip and it becomes wildly complex, okay? So this is very provocative,

01:52:34.320 | very interesting. It's also not how physics works at all because as we said, physics conserves

01:52:41.680 | information. You can go forward or backwards. These cellular automata do not. They're not

01:52:46.800 | reversible in any sense. You've built in an arrow of time. You have a starting point and then you

01:52:52.400 | evolve. So what I'm interested in is seeing how in the real world with the real laws of physics

01:52:58.000 | and underlying reversibility, but macroscopic irreversibility from entropy in the arrow

01:53:03.680 | of time, et cetera, how does that lead to complexity? I think that that's an answerable

01:53:08.240 | question. I don't think that cellular automata are really helping us in that one.

01:53:10.960 | >> So what is in that, what is the landscape of entropy in the universe look like?

01:53:17.600 | >> Well, entropy is hard to localize. It's a property of systems, not of parts of systems,

01:53:24.080 | right? Having said that, we can do approximate answers to the question. The answer is black

01:53:30.480 | holes are huge in entropy. Most of, let's put it this way, the whole observable universe that we're

01:53:37.840 | in had a certain amount of entropy before stars and planets and black holes started to form,

01:53:45.120 | 10 to the 88th. I can even tell you the number, okay? The single black hole at the center of our

01:53:51.040 | galaxy has entropy, 10 to the 90th. The single black hole at the center of our galaxy has more

01:53:56.320 | entropy than the whole universe used to have not too long ago. So most of the entropy in the

01:54:02.000 | universe today is in the form of black holes. >> Okay, that's fascinating, first of all.

01:54:07.040 | But second of all, if we take black holes away, what are the different interesting perturbations

01:54:12.400 | in entropy across space? Where do we earthlings fit into that?

01:54:18.000 | >> The interesting thing to me is that if you start with a system that is isolated from the

01:54:25.040 | rest of the universe, and you start it at low entropy, there's almost a theorem that says if

01:54:31.600 | you're very, very, very low entropy, then the system looks pretty simple. Because low entropy

01:54:37.280 | means there's only a small number of ways that you can rearrange the parts to look like that.

01:54:42.400 | So if there's not that many ways, the answer is going to look simple. But there's also almost

01:54:46.880 | a theorem that says when you're at maximum entropy, the system is going to look simple,

01:54:51.200 | because it's all smeared out. If it had interesting structure, then it would be complicated, right?

01:54:56.560 | So entropy in this isolated system only goes up. That's the second law of thermodynamics.

01:55:02.000 | But complexity starts low, goes up, and then goes down again.

01:55:07.120 | Sometimes people mistakenly think that complexity or life or whatever is fighting against the second

01:55:16.240 | law of thermodynamics, fighting against the increase of entropy. That is precisely the

01:55:20.320 | wrong way to think about it. We are surfers riding the wave of increasing entropy. We rely

01:55:27.440 | on increasing entropy to survive. That is part of what makes us special. This table

01:55:32.640 | maintains its stability mechanically, by which I mean there's molecules, they have forces on

01:55:40.640 | each other, and it holds up. You and I aren't like that. We maintain our stability dynamically

01:55:47.920 | by ingesting food, fuel, right? Food and water and air and so forth,

01:55:52.480 | burning it, increasing its entropy. We are non-equilibrium quasi-steady-state systems.

01:55:59.120 | We are using the fuel the universe gives us in the form of low entropy energy to maintain our

01:56:05.200 | stability. - I just wonder what that mechanism of surfing looks like. I mean, that's where,

01:56:10.800 | first of all, I mean, one question to ask, do you think it's possible to have a kind of science of

01:56:16.160 | complexity where you have very precise ways or clearly defined ways of measuring complexity?

01:56:24.240 | - I think it is, and I think we don't. It's possible to have it, I don't think we yet have it.

01:56:29.680 | In part because complexity is not a univalent thing. There's different ideas that go under

01:56:36.320 | the rubric of complexity. One version is just Kamalgal of complexity, right? If you have a

01:56:42.240 | configuration or a string of numbers or whatever, can you compress it so that you have a small

01:56:48.960 | program that will help with that? That's Kamalgal of complexity. But that's the complexity of a

01:56:54.000 | string of numbers, okay? It's not like the complexity of a problem, right? Computational

01:57:00.000 | complexity, the traveling salesman problem or factoring large numbers. That's a whole different

01:57:04.160 | kind of question that is also about complexity. So we don't have a sort of unified view of it.

01:57:09.440 | - Do you think it's possible to have a complexity of a physical system?

01:57:13.200 | - Yeah, absolutely.

01:57:13.920 | - In the same way we do entropy?

01:57:15.520 | - Yeah.

01:57:16.080 | - You think that's a Sean Carroll paper or what?

01:57:19.040 | - We're working on various things. The glib thing that I'm trying to work on right now with a

01:57:25.840 | student is complexogenesis. How does complexity come to be if all the universe is doing is moving

01:57:31.840 | from low entropy to high entropy?

01:57:32.960 | - It's a sexy name.

01:57:34.320 | - It's a good name. Yeah, I like the name. I just got to write the paper.

01:57:36.960 | - Sometimes a name arose by any other name. In which context the birth of complexity are you

01:57:48.560 | most interested in?

01:57:49.280 | - Well, I think it comes in stages, right? So I think that if you go from the, I'm again a

01:57:55.440 | physicist. So biologists studying evolution will talk about how complexity evolves all the time.

01:58:02.160 | The complexity of the genome, the complexity of our physiology, right? But they take for granted

01:58:07.200 | that life already existed and entropy is increasing and so forth. I want to go back to

01:58:13.200 | the beginning and say the early universe was simple and low entropy. And entropy increases

01:58:18.400 | with time and the universe sort of differentiates and becomes more complex. But that statement,

01:58:24.640 | which is indisputably true, has different meanings because complexity has different

01:58:29.520 | meanings. So sort of the most basic primal version of complexity is what you might think

01:58:36.080 | of as configurational complexity. That's what Komolgorov gets at. How much information do you

01:58:41.200 | need to specify the configuration of the system? Then there's a whole other step where subsystems

01:58:48.400 | of the universe start burning fuel, right? So in many ways, a planet and a star are not that

01:58:55.920 | different in configurational complexity. They're both spheres with density high at the middle and

01:59:00.880 | getting less as you go out. But there's something fundamentally different because the star only

01:59:05.280 | survives as long as it has fuel, right? I mean, then it turns into a brown dwarf or a white dwarf

01:59:09.040 | or whatever. But as a star, as a main sequence star, it is an out of equilibrium system.

01:59:14.880 | But it's more or less static, right? Like if I spill the coffee mug and it falls in the process

01:59:20.880 | of falling, it's out of equilibrium, but it's also changing all the time. A specific kind of

01:59:26.320 | system is where it looks sort of macroscopically stationary like a star, but underneath the hood,

01:59:34.080 | it's burning fuel to beat the band in order to maintain that stability. So as stars form,

01:59:39.840 | that's a different kind of complexity that comes to be. Then there's another kind of complexity

01:59:44.800 | that comes to be roughly speaking at the origin of life because that's where you have information

01:59:52.000 | really being gathered and utilized by subsystems of the universe. And then arguably, there's any

01:59:58.080 | number of stages past that. I mean, one of the most obvious ones to me is we talk about simulation

02:00:04.560 | theory, but you and I run simulations in our heads. They're just not that good, but we imagine

02:00:09.680 | different hypothetical futures, right? Bacteria don't do that. So that's the kind of information

02:00:15.040 | processing that is a form of complexity. So I would like to understand all these stages and

02:00:19.040 | how they fit together. Yeah, imagination.

02:00:21.760 | Yeah. Mental time travel.

02:00:23.440 | Yeah. The things going on in my head when I'm imagining worlds are super compressed

02:00:29.840 | representations of those worlds, but they get to the essence of them. And maybe it's possible with

02:00:35.440 | non-human computing type devices to do those kinds of simulations in more and more compressed ways.

02:00:41.200 | There's an argument to be made that literally what separates human beings from other

02:00:46.240 | species on Earth is our ability to imagine counterfactual hypothetical futures.

02:00:51.680 | Yeah. I mean, that's one of the big features. I don't know if it's a-

02:00:59.440 | Everyone has their own favorite little feature, but that's why I said there's an argument to

02:01:02.640 | be made. I did a podcast episode on it with Adam Bulley. It developed slowly. I did a different

02:01:08.240 | podcast. Sorry to keep mentioning podcast episodes I did, but Malcolm McIver, who is an engineer at

02:01:12.400 | Northwestern, has a theory about one of the major stages in evolution is when fish first climbed on

02:01:19.120 | the land. And I mean, of course, that is a major stage of evolution, but in particular, there's

02:01:23.280 | a cognitive shift. Because when you're a fish swimming under the water, the attenuation length

02:01:30.400 | of light in water is not that long. You can't see kilometers away. You can see meters away.

02:01:36.240 | And you're moving at meters per second. So all of the evolutionary optimization is make all of your

02:01:43.680 | decisions on a time scale of less than a second. When you see something new, you have to make a

02:01:48.080 | rapid fire decision what to do about it. As soon as you climb onto land, you can essentially see

02:01:53.680 | forever, right? You can see stars in the sky. So now a whole new mode of reasoning opens up where

02:02:02.160 | you see something far away. And rather than saying, "Look up, people. I see this. I react,"

02:02:07.280 | you can say, "Okay, I see that thing. What if I did this? What if I did that? What if I did

02:02:12.800 | something different?" And that's the birth of imagination eventually. >> You've been critical

02:02:18.000 | on panpsychism. >> Yes, you've noticed that, right. >> Can you make the case for panpsychism

02:02:24.320 | and against it? So panpsychism is the idea that consciousness permeates all matter. Maybe

02:02:30.160 | it's the fundamental force or physics of the fabric of the universe. >> Panpsychism. Thought

02:02:41.600 | everywhere. Consciousness everywhere, right? >> It's to a point of entertainment. The idea

02:02:47.680 | frustrates you, which sort of as a fan is wonderful to watch. And you've had great episodes

02:02:54.000 | with panpsychists on your podcast where you go at it. >> I had David Chalmers, who's one of the

02:03:00.240 | world's great philosophers, and he is panpsychism curious. He doesn't commit to anything, but he's

02:03:07.680 | certainly willing to entertain it. Philip Goff, who I've had, who's a great guy, but he is devoted

02:03:13.840 | to panpsychism. In fact, he is almost single-handedly responsible for the upsurge of interest in

02:03:19.600 | panpsychism in the popular imagination. And the argument for it is supposed to be that there is

02:03:25.840 | something fundamentally uncapturable about conscious awareness by physical behavior of atoms

02:03:33.120 | and molecules. So the panpsychist will say, "Look, you can tell me maybe someday through advances of

02:03:38.960 | neuroscience and what have you exactly what happens in your brain and how that translates into thought

02:03:46.080 | and speech and action. What you can't tell me is what it is like to be me. You can't tell me what

02:03:55.120 | I am experiencing when I see something that is red or that tastes something that is sweet.

02:04:02.080 | You can tell me what neurons fire, but you can't tell me what I'm experiencing. That first-person

02:04:07.280 | inner subjective experience is simply not capturable by physics. And therefore,

02:04:15.600 | this is an old argument, of course, but then the therefore is supposed to be,

02:04:19.840 | I need something that is not contained within physics to account for that. And I'm just going

02:04:25.840 | to call it mind. We don't know what it is yet. We're going to call it mind. And it has to be

02:04:30.160 | separate from physics. And then there's two ways to go. If you buy that much, you can either say,

02:04:36.400 | "Okay, I'm going to be a dualist. I'm going to believe that there's matter and mind and they

02:04:40.480 | are separate from each other and they are interacting somehow." Or that's a little bit

02:04:45.920 | complicated and sketchy as far as physics is going to go. So I'm going to believe in mind,

02:04:49.600 | but I'm going to put it prior to matter. I'm going to believe that mind comes first

02:04:54.160 | and that consciousness is the fundamental aspect of reality and everything else,

02:04:59.200 | including matter and physics, comes from it. That would be at least as simple as physics comes first.

02:05:06.160 | Now, the physicalist, such as myself, will say, "I don't have any problem explaining

02:05:14.560 | what it's like to be you or what you experience when you see red. It's a certain way of talking

02:05:20.080 | about the atoms and the neurons, et cetera, that make up you. Just like the hardness or the

02:05:27.120 | brownness of this table, these are words that we attach to certain underlying configurations

02:05:33.360 | of ordinary physical matter. Likewise, sadness and redness or whatever are words that we attach

02:05:40.080 | to you to describe what you're doing." When it comes to consciousness in general, I'm very quick

02:05:47.360 | to say I do not claim to have any special insight on how consciousness works other than I see no

02:05:55.520 | reason to change the laws of physics to account for it. - If you don't have to change the laws

02:05:59.360 | of physics, where do you think it emerges from? Is consciousness an illusion that's almost like

02:06:05.520 | a shorthand that we humans use to describe a certain kind of feeling we have when interacting

02:06:09.920 | with the world? Or is there some big leap that happens at some stage? - I almost never use the

02:06:16.720 | word illusion. Illusion means that there's something that you think you're perceiving

02:06:21.040 | that is actually not there. Like an oasis in the desert is an illusion. It has no causal efficacy.

02:06:28.240 | If you walk up to where the oasis is supposed to be, you'll say you were wrong about it being there.

02:06:33.040 | That's different than something being emergent or non-fundamental but also real. Like this table is

02:06:37.920 | real, even though I know it's made of atoms. That doesn't remove the realness from the table. I

02:06:43.120 | think that consciousness and free will and things like that are just as real in tables and chairs.

02:06:47.360 | - Oasis in the desert does have causal efficacy in that you're thirsty.

02:06:51.920 | - Incorrect causal efficacy. It leads to draw incorrect conclusions about the world.

02:06:55.760 | - Sure, but imagining a thing can sometimes bring it to reality, as we've seen, and that has a kind

02:07:03.760 | of causal efficacy. - Sure, but your understanding of the world

02:07:10.000 | in a way that gives you power over it and influence over it is decreased rather than

02:07:14.720 | increased by believing in that oasis. That is not true about consciousness or this table.

02:07:19.680 | - You don't think you can increase the chance of a thing existing by imagining it existing?

02:07:27.200 | - Unless you build it or make it. - No, that's what I mean. Like

02:07:32.800 | imagining humans can fly. - That's different than imagining

02:07:36.560 | that humans are flying. - Right.

02:07:39.280 | - In terms of counterfactuals in the future, absolutely. Imagination is crucially important.

02:07:44.640 | But that's not an illusion. That's just imagination. - Oh, okay, so the possibility of the

02:07:49.440 | future versus what reality is. I mean, the future is a concept, so you can, well, and time.

02:07:58.400 | Time is just a concept, so you can play with that. But yes, reality.

02:08:08.800 | So to you, so for example, I love asking this. So Donald Hoffman

02:08:14.560 | thinks that the entirety of the conversation we've been having about space-time

02:08:22.640 | is an illusion. Is it possible for you to steelman the case for that?

02:08:26.400 | Can you make the case for and against reality, as I think he writes, that the laws of physics

02:08:35.600 | as we know them with space-time is a kind of interface to a much deeper thing that we don't

02:08:40.480 | at all understand and that we're fooling ourselves by constructing this world?

02:08:44.320 | - Well, I think there's like part of that idea that is perfectly respectable and part of it

02:08:48.880 | that is perfectly nonsensical. And I'm not even gonna try to steelman the nonsensical part.

02:08:53.120 | The real part to me is what is called structural realism. So we don't know what the world is

02:09:02.800 | at a deep fundamental level, right? Let's put ourselves in the minds of people living 200

02:09:09.280 | years ago. They didn't know about quantum mechanics, they didn't know about relativity.

02:09:12.640 | That doesn't mean they were wrong about the universe that they understood. They had Newton's

02:09:18.160 | laws, right? They could predict what time the Sun was gonna rise perfectly well. In the progress of

02:09:24.240 | science, the words that would be used to give the most fundamental description of how you

02:09:32.400 | were predicting the Sun would rise changed because now you have curved space-time and

02:09:37.360 | things like that, right? And you didn't have any of those words 200 years ago.

02:09:40.720 | But the prediction is the same. Why? Because that prediction, independent of what we thought

02:09:47.040 | the fundamental ontology was, the prediction pointed to something true about our understanding

02:09:54.640 | of reality. To call it an illusion is just wrong, I think. We might not know what the best,

02:10:01.280 | most comprehensive way of stating it is, but it's still true.

02:10:05.200 | >> Is it true in the way, for example, belief in God is true? Because for most of human history,

02:10:12.800 | people have believed in a god or multiple gods, and that seemed very true to them.

02:10:21.600 | As an explanation for the way the world is, some of the deeper questions about life itself,

02:10:29.920 | and the human condition, and why certain things happen, that was a good explainer.

02:10:34.320 | So to you, that's not an illusion? >> No, I think that was completely

02:10:41.280 | an illusion. I think it was a very, very reasonable illusion to be under. There are

02:10:44.880 | illusions. There are substantive claims about the world that go beyond predictions that we can make

02:10:52.000 | and verify, which later turn out to be wrong. And the existence of God was one of them.

02:10:58.320 | If those people at that time had abandoned their belief in God and replaced it with a

02:11:03.040 | mechanistic universe, they would have done just as well at understanding things.

02:11:06.800 | Again, because there's so many things they didn't understand, it was very reasonable for them to

02:11:12.960 | have that belief. It wasn't that they were dummies or anything like that. But that is,

02:11:17.200 | as we understand the universe better and better, some things stick with us, some things get replaced.

02:11:22.720 | >> So like you said, you're a believer of the mechanistic universe. You're a naturalist,

02:11:31.120 | and as you've described, a poetic naturalist. >> That's right.

02:11:35.200 | >> What's the word poetic? What is naturalism, and what is poetic naturalism?

02:11:39.520 | >> Naturalism is just the idea that all that exists is the natural world.

02:11:42.880 | There's no supernatural world. You can have arguments about what that means,

02:11:48.240 | but I would claim that the argument should be about what the word supernatural means,

02:11:53.040 | not the word natural. The natural world is the world that we learn about by doing science.

02:11:56.560 | The poetic part means that you shouldn't be too,

02:12:00.560 | I want to say, fundamentalist about what the natural world is. As we went from

02:12:07.760 | Newtonian space-time to Einsteinian space-time, something is maintained there. There is a

02:12:16.240 | different story that we can tell about the world, and that story in the Newtonian regime,

02:12:22.160 | if you want to fly a rocket to the moon, you don't use general relativity. You use Newtonian

02:12:25.840 | mechanics. That story works perfectly well. The poetic aspect of the story is that there are many

02:12:31.360 | ways of talking about the natural world, and as long as those ways latch on to something real and

02:12:38.560 | causally efficacious about the functioning of the world, then we attribute some reality and truth

02:12:43.600 | to them. >> So the poetic really looks at the,

02:12:46.240 | let's say, the pothead questions at the edge of science. It's more open to them.

02:12:52.320 | >> It's doing double duty a little bit, so that's why it's confusing. The more obvious respectable

02:12:57.280 | duty it's doing is that tables are real. Even though you know that it's really a quantum field

02:13:03.920 | theory wave function, tables are still real. They're a different way of talking about

02:13:09.120 | the underlying deeper reality of it. The other duty it's doing is that we move beyond purely

02:13:14.880 | descriptive vocabularies for discussing the universe onto normative and prescriptive and

02:13:21.040 | judgmental ways of talking about the universe. This painting is beautiful, that one is ugly.

02:13:26.320 | This action is morally right, that one is morally wrong. These are also ways of talking about the

02:13:32.000 | universe. They are not fixed by the phenomena. They are not determined by our observations.

02:13:38.080 | They cannot be ruled out by a crucial experiment, but they're still valid. They might not be

02:13:42.720 | universal, they might be subjective, but they're not arbitrary, and they do have a role in describing

02:13:48.800 | how the world works. >> So you don't think it's possible to

02:13:51.360 | construct experiments that explore the realms of morality and even meaning? So those are subjective?

02:14:01.520 | >> Yeah, they're human, they're personal. >> But do you think that's just because we

02:14:06.160 | don't have a, the tools of science have not expanded enough to incorporate the human experience?

02:14:12.640 | >> No, I don't think that's what it is. I think that what we mean by aesthetics or morality

02:14:17.520 | are we're attaching categories, properties, to things that happen in the physical world.

02:14:24.320 | And there is always going to be some subjectivity to our attachment and how we do that, and that's

02:14:28.480 | okay. And the faster we recognize that and deal with it, the better off we'll be.

02:14:31.440 | >> But if we deeply and fully understand the function of the human mind, we'll be able to

02:14:38.000 | incorporate that? >> No, that will absolutely be helpful in

02:14:41.760 | explaining why certain people have certain moral beliefs. It won't justify those beliefs as right

02:14:47.280 | or wrong. >> Do you think it's possible to have a kind

02:14:49.200 | of general relativity, but that includes the observer effect where the human mind is the

02:14:55.600 | observer? Sort of like how we morph, in the same way gravity morphs space-time, how does the human

02:15:04.400 | mind morph reality and have a very thorough theory of how that morphing actually happens?

02:15:14.400 | >> That's a very plot-head question, Lex. >> But it is possible.

02:15:18.480 | >> But the answer is yes. I think that there's no, I think that we're part of the physical world,

02:15:25.040 | and the natural world. Physicalism would have been just as good a word to use as naturalism,

02:15:31.200 | maybe even a more accurate word, but it's a little bit more off-putting, so I do want a

02:15:35.120 | snappier, more attractive label than physicalism. >> Are there limits to science?

02:15:41.760 | >> Sure. We just talked about one, right? Science can't tell you right from wrong.

02:15:45.280 | You need science to implement your ideas about right and wrong. If you are functioning on the

02:15:53.120 | basis of an incorrect view of how the world works, you might very well think you're doing right but

02:15:57.840 | actually be doing wrong. But all the science in the world won't tell you which action is right

02:16:02.720 | and which action is wrong. >> You know, dictators and people in

02:16:06.960 | power will sometimes use science as an authority to convince you what's right and wrong. Studying

02:16:14.240 | Nazi science is fascinating. >> Yeah, but there's an instrumentalist

02:16:17.920 | view here. You have to first decide what your goals are, and then science can help you achieve

02:16:22.480 | those goals. If your goals are horrible, science has no problem helping you achieve them. Science

02:16:28.240 | is happy to help out. >> Let me ask you about the method behind

02:16:31.200 | the madness on several aspects of your life. So you mentioned that your approach to writing for

02:16:37.360 | research and writing popular books, how do you find the time of the day? Like, what's the day

02:16:42.560 | in the life of Sean Carroll looks like? >> Very unclear how I find the time.

02:16:45.520 | >> So you don't have a thing where in the morning you're like, you try to fight for

02:16:49.760 | two hours somewhere? >> I don't. I'm really terrible at that.

02:16:52.800 | My strategy for finding time is just to ignore interruptions and emails. But it's a different

02:16:59.600 | time every day, some days it never happens, some weeks it never happens.

02:17:03.520 | >> Well, really, you're able to pull it off because you're extremely prolific. So you're

02:17:06.800 | able to have days where you don't write and still write the next day. Oh, wow. That's a rare thing,

02:17:14.400 | right? A lot of prolific writers will carve out two hours because otherwise it just disappears.

02:17:21.280 | >> Right. No, I get that. Yeah, I do. And yeah, it's just like everyone has their foibles or

02:17:30.240 | whatever. So I'm not able to do that, therefore I have to just figure it out on the fly.

02:17:37.280 | >> And what's the actual process look like when you're writing popular stuff? You get

02:17:41.360 | behind a computer? >> Yeah, get behind a computer. And my way

02:17:44.720 | of doing it, so my wife, Jennifer, is a science writer. But it's interesting because our techniques

02:17:50.400 | are entirely different. She will think about something, but then she will free write. She'll

02:17:54.720 | just sit at a computer and write. Like, I think this, I think this. And then that will be vastly

02:18:00.320 | compressed, edited, rewritten or whatever until the final thing happens. I will just sit there

02:18:06.720 | silently thinking for a very long time, and then I will write what is almost the final draft. So a

02:18:12.320 | lot of it happens. There might be some scribbles for an outline or something like that, but a lot

02:18:16.320 | of it is in my brain before it's on the page. >> So that's the case for the biggest ideas in

02:18:20.080 | the universe, the quanta book and the space-time motion book?

02:18:22.960 | >> Yeah, quanta and fields, which is actually mostly about quantum field theory and particle

02:18:27.040 | physics. That's coming out in May. And that is, I'm letting people in on things that no other book

02:18:35.760 | lets them in on. So I hope it's worth it. It's a challenge because there's a lot of equations.

02:18:39.920 | >> I mean, you did the same thing with space-time motion. You did something quite interesting,

02:18:44.320 | which is like you made the equation the centerpiece of a book.

02:18:48.240 | >> Right. There's a lot of equations. Book two goes further in those directions than book one did.

02:18:56.400 | So it's more cool stuff. It's also more mind-bending. It's more of a challenge.

02:19:03.440 | Book three that I'm writing right now is called Complexity and Emergence.

02:19:08.400 | >> Oh, wow. >> And that'll be the final

02:19:10.400 | part of the trilogy. >> Oh, that's fascinating.

02:19:12.800 | So there's a lot of probably ideas there. I mean, that's a real cutting edge.

02:19:16.880 | >> Well, but I'm not trying to be cutting edge. In other words, I'm not trying to speculate in

02:19:23.200 | these books. Obviously, in other books, I've been very free about speculating. But the point of

02:19:28.080 | these books is to say things that 500 years from now will still be true. And so there are some

02:19:33.760 | things we know about complexity and emergence, and I want to focus on those. And I will mention,

02:19:38.560 | I'm happy to say, this is something that needs to be speculated about, but I won't pretend to

02:19:42.640 | be telling you what one is the right one. >> You somehow found the balance between

02:19:46.240 | the rigor of mathematics and still accessible, which is interesting.

02:19:49.680 | >> I try. I mean, look, these three books, the biggest ideas books, are absolutely an experiment.

02:19:56.160 | They're going to appeal to a smaller audience than other books will. But that audience should

02:20:02.560 | love them. My 16-year-old self would have been so happy to get these books, I can't tell you.

02:20:07.440 | >> Yeah, in terms of looking back in history, those books, the trilogy would be truly special

02:20:12.640 | in that way. Worked for Lord of the Rings, so I figured, why not me? You and Tolkien are just-

02:20:17.760 | >> Yeah, different styles, different topics. >> Same ultimate reality.

02:20:22.880 | >> Like we mentioned, Mindscape Podcast, I love it. You interview a huge variety of experts

02:20:31.280 | from all kinds of fields, so just several questions I want to ask. How do you prepare?

02:20:36.400 | Like, how do you prepare to have a good conversation? How do you prepare in a way that

02:20:42.160 | satisfies, makes your own curious mind happy, all that kind of stuff?

02:20:46.160 | >> Yeah, no, these are great questions, and I've sort of struggled and changed my

02:20:50.160 | techniques over the years. It's over five-year-old podcast, might be approaching six years old now.

02:20:54.960 | I started out over-preparing when I first started. I had a journey that I was going to go down.

02:21:03.440 | Many of the people I talk to are academics or thinkers who write books, so they have a story

02:21:08.560 | to tell. I could just say, "Okay, give me your lecture," and then an hour later, stop, right?

02:21:15.360 | So the mistake is to sort of anticipate what the lecture would be and to ask the leading questions

02:21:20.720 | that would pull it out of them. What I do now is much more, here are the points, here are like the

02:21:27.200 | big questions that I'm interested in, and so I have a much sketchier outline to start, and then

02:21:34.240 | try to make it more of a real conversation. I'm helped by the fact that it is not my day job,

02:21:41.360 | so I strictly limit myself to one day of my life per podcast episode, on average. Some days take

02:21:50.320 | more, and that includes not just doing the research, but inviting the guests, recording it,

02:21:55.120 | editing it, publishing it. So I need to be very, very efficient at that, yeah.

02:21:59.760 | >> You enforce constraints for yourself in which creativity can emerge.

02:22:03.280 | >> That's right, that's right. And look, sometimes if I'm interviewing a theoretical physicist,

02:22:09.920 | I can just go in. And when I'm interviewing an economist or a historian, I have to do a lot of

02:22:15.280 | work. >> Do you ever find yourself

02:22:16.880 | getting lost in rabbit holes that serve no purpose except satisfying your own curiosity,

02:22:23.520 | and then potentially expanding the range of things you know that can help your actual

02:22:28.880 | work and research and writing? >> Yes, on both counts. You know, I do,

02:22:33.360 | some people have so many things to talk about that you don't know where to start or finish,

02:22:39.200 | right? Others have a message. And one of the things I discovered over the course of these years

02:22:45.280 | is the correlation with age. Like, there are brilliant people, and I try very hard on the

02:22:50.800 | podcast to sort of get all sorts of people, right, different ages and things like that.

02:22:55.600 | And bless their hearts, the most brilliant young people are not as practiced at wandering past

02:23:03.840 | their literal research, right? They have less mastery over the field as a whole,

02:23:09.440 | much less how to talk about it. Whereas certain older people just like have their patent answers,

02:23:14.400 | and that's kind of boring, right? So you want somewhere in between, you know, the ideal person

02:23:18.320 | has a broad enough scope that they can wander outside their specific papers they've written,

02:23:25.360 | but they're not overly practiced, so they're just giving you their canned answers.

02:23:28.640 | >> I feel like there's like a connection to the metaphor of entropy and complexity,

02:23:32.400 | as you said there. You also do incredible AMAs, and people should sign up to your Patreon because

02:23:40.480 | you can get to ask questions, Sean Carroll. For several hours, you just answer in fascinating

02:23:49.360 | ways some really interesting questions. Is there something you could say about the process of

02:23:54.800 | finding the answers to those? >> That's a great one. Again,

02:23:58.240 | it's evolved over time. Yeah, so the Ask Me Anything episodes were first, when I started

02:24:04.880 | doing them, they were only for Patreon subscribers to both listen to and to ask the questions. But

02:24:10.560 | then I actually asked my Patreon subscribers, "Would you like me to release them publicly?"

02:24:15.600 | And they overwhelmingly voted yes, so I do that. So the Patreon supporters ask the questions,

02:24:20.080 | everyone can listen. And also at some point, I really used to try to answer every question,

02:24:26.240 | but now there's just too many, so I have to pick, and that's fraught with peril. And my personal

02:24:32.160 | standard for picking questions to answer is what are the ones I think I have interesting answers

02:24:37.440 | to give for, right? So that both means if it's kind of the same old question about special

02:24:43.520 | relativity that I've gotten 100 times before, I'm not going to answer it, because you can just

02:24:47.920 | Google that, it's easier. There are some very clear attempts to ask an interesting question

02:24:56.800 | that honestly just I don't have an answer to. Like, "I read this science fiction novel,

02:25:01.680 | what do you think about it?" I'm like, "Well, I haven't read it, so I can't help you there."

02:25:05.520 | "What's your favorite color?" I could tell you what it is, but it's not that interesting.

02:25:10.480 | And so I try to make it a mix. It's not all physics questions, not all philosophy questions.

02:25:18.160 | I will talk about food or movies or politics or religion if that's what people want. I keep

02:25:23.280 | suggesting that people ask me for relationship advice, but they never do.

02:25:26.480 | >> Yeah, I don't think I've heard one. >> I'm willing to do it, but I'm a little

02:25:33.440 | reluctant because I don't actually like giving advice, but I'm happy to talk about those topics.

02:25:39.280 | I want to give several hours of talking, and I want to try to say things that I haven't said

02:25:46.960 | before and keep it interesting, keep it rolling. If you don't like this question, wait for the

02:25:50.000 | next one. >> What are some of the

02:25:51.040 | harder questions you've gotten, do you remember? What kinds of questions are difficult for you?

02:25:55.520 | >> Rarely, but occasionally, people will ask me a super insightful philosophy question.

02:26:02.000 | Like, I hadn't thought of things in exactly that way, and I try to recognize that.

02:26:08.800 | A lot of times, it's the opposite, where it's like, OK, you're clearly confused,

02:26:16.560 | and I'm going to try to explain the question you should have asked.

02:26:19.920 | >> I love those. Yeah, why that's the wrong question or that kind of stuff, that's great.

02:26:23.600 | >> Right. >> That's great.

02:26:24.720 | >> But the hard questions, I don't know. I don't actually answer personal questions very much.

02:26:30.720 | The most personal I will get are questions like, "What do you think of Baltimore?"

02:26:34.080 | Right? That much I can talk about. Or, "How are your cats doing?" Happy to talk about the cats

02:26:38.080 | in infinite detail. But very personal questions I don't get into.

02:26:42.480 | >> But you even touch politics and stuff like this.

02:26:45.520 | >> Yeah, no, very happy to talk about politics. I try to be clear on what is professional

02:26:52.160 | expertise, what is just me babbling, what is my level of credence in different things,

02:26:55.840 | where you're allowed to disagree, whether if you disagree, you're just wrong.

02:27:00.000 | And people can disagree with that also. But I do think, and I'm happy to go out on a limb

02:27:08.000 | a little bit. I'm happy to say, "Look, I don't know, but here's my guess." Right?

02:27:11.600 | I just did a whole solo podcast, which was exactly that. And it's interesting, some people are like,

02:27:16.800 | "Oh, this was great." And there's a whole bunch of people like, "Why are you talking

02:27:19.760 | about this thing that you are not the world's expert in?" So, you know?

02:27:23.200 | >> Well, I love the actual dance between humility and having a strong opinion on stuff,

02:27:28.560 | which is a great, it's a fascinating dance to pull off. And I guess the way to do that is to just

02:27:34.880 | expand into all kinds of topics and play with ideas and then change your mind and all that

02:27:40.000 | kind of stuff. >> Yeah, it's interesting because when people react against you by saying, "You are

02:27:48.240 | being arrogant about this," 99.999% of the time, all they mean is, "I disagree." That's all they

02:27:56.000 | really mean, right? Like, at a very basic level, people will accuse atheists of being arrogant.

02:28:05.280 | And I'm like, "You think God exists and loves you, and you're telling me that I'm arrogant."

02:28:10.320 | I think that all of this is to say, just advice. When you disagree with somebody,

02:28:18.320 | try to specify the substantive disagreement. Try not to psychologize them, right? Try to say, "Oh,

02:28:23.840 | you're saying this because of this." Maybe it's true, maybe you're right. But if you had an actual

02:28:29.440 | response to what they were saying, that would be much more interesting. >> Yeah, I think, I wonder

02:28:34.080 | why it's difficult for people to say or to imply, "I respect you, I like you, but I disagree on this,

02:28:42.000 | and here's why I disagree." I wonder why they go to this place of like, "Well, you're an idiot," or

02:28:51.440 | "You're egotistical," or "You're confused," or "You're naive," or all the kinds of words,

02:29:01.200 | as opposed to like, "I respect you as a fellow human being exploring the world of mysteries

02:29:06.240 | all around us, and I disagree." >> I will complicate the question even more,

02:29:10.880 | because there's some people I don't respect or like. And I once wrote a blog post, I think it

02:29:17.360 | was called The Grid of Disputation. And I had a two-by-two grid, and it's, "Are you someone I

02:29:23.440 | agree with or disagree with? Are you someone who I respect or don't?" And all four quadrants are

02:29:31.680 | very populated. And so what that means is, there are people who I like and I disagree with, and

02:29:38.960 | there are people who agree with me and I have no respect for at all, the embarrassing allies

02:29:43.120 | quadrant. That was everyone's favorite. And I just think being honest, trying to be honest about

02:29:50.240 | where people are, but if you actually want to move a conversation forward, forget about whether you

02:29:55.760 | like or don't like somebody. Explain the disagreement, explain the agreement. But

02:30:00.160 | you're absolutely right. I completely agree, as a society, we are not very good at disagreeing.

02:30:04.560 | We instantly go to the insults. >> Yeah, and I mean, even on a deeper level,

02:30:08.480 | I think at some deep level, I respect and love the humanity in the other person.

02:30:18.240 | >> Yep. >> You said that general

02:30:23.840 | relativity is the most beautiful theory ever? >> So far.

02:30:27.120 | >> What do you find beautiful about it? >> Let's put it this way. When I teach courses,

02:30:34.480 | there's no more satisfying subject to teach than general relativity. And the reason why is because

02:30:40.880 | it starts from very clear, precisely articulated assumptions, and it goes so far. And when I give

02:30:50.480 | my talk, you can find it online, I'm probably not going to give it again. The book, one of the

02:30:54.000 | biggest ideas, talk, was building up from, you don't know any math or physics, an hour later,

02:31:01.360 | you know Einstein's equation for general relativity. And the punchline is, the equation

02:31:07.840 | is much smarter than Albert Einstein. Because Albert Einstein did not know about the Big Bang,

02:31:14.000 | he didn't know about gravitational waves, he didn't know about black holes, but his equation did.

02:31:18.800 | And that's a miraculous aspect of science more generally, but general relativity is

02:31:26.960 | where it manifests itself in the most absolutely obvious way. >> A human question,

02:31:32.800 | what do you think of the fact that Einstein didn't get the Nobel Prize for general relativity?

02:31:39.920 | >> Tragedy. He should have gotten maybe four Nobel Prizes, honestly.

02:31:45.360 | >> That one. >> He certainly should have got,

02:31:48.320 | the photoelectric effect was 100% worth the Nobel Prize, because, and people don't quite get this,

02:31:53.840 | who cares about the photoelectric effect? That's like this very minor effect. The point is his

02:31:57.920 | explanation for the photoelectric effect invented something called the photon. That's worth the

02:32:04.800 | Nobel Prize. Max Planck gets credit for this in 1900, explaining black-body radiation by saying

02:32:12.880 | that when a little electron is jiggling in an object at some temperature, gives off radiation

02:32:20.480 | in discrete chunks rather than continuously. He didn't quite say that's because radiation

02:32:28.160 | is discrete chunks, right? It's like having a coffee maker that makes one cup of coffee at

02:32:32.800 | a time. It doesn't mean that liquid comes in one cup quanta, right? It's just that you are

02:32:37.360 | dispensing it like that. It was Einstein in 1905 who said light is quanta, and that was a radical

02:32:43.920 | thing. So that clearly, that was not a mistake, but also special relativity clearly deserved the

02:32:49.360 | Nobel Prize, and general relativity clearly deserved the Nobel Prize. Not only were they

02:32:54.080 | brilliant, but they were experimentally verified, like everything you want.

02:32:56.880 | - So separately, you think?

02:32:58.160 | - Yeah, yeah, absolutely.

02:32:59.840 | - Oh, humans, whatever the explanation there.

02:33:04.560 | - Edwin Hubble never won the Nobel Prize for finding the universe was expanding.

02:33:09.600 | - Yeah, and even the fact that we give prizes is almost kind of silly, and we limit the number

02:33:14.960 | of people that get the prize and all that.

02:33:16.960 | - I think the Nobel Prize has enormous problems. I think it's probably a net good for the world

02:33:23.680 | because it brings attention to good science. I think it's probably a net negative for science

02:33:29.680 | because it makes people want to win the Nobel Prize.

02:33:32.560 | - Yeah, there's a lot of fascinating human stories underneath it all. Science is its own thing,

02:33:39.360 | but it's also a collection of humans, and it's a beautiful collection. There's tension,

02:33:43.520 | there's competition, there's jealousy, but there's also great collaborations and all that kind of

02:33:50.080 | stuff. Daniel Kahneman, who recently passed, is one of the great stories of collaboration in science.

02:34:00.320 | So all of it, all of it, that's what humans do. And Sean, thank you for being the person that

02:34:07.600 | makes us celebrate science and fall in love with all of these beautiful ideas in science,

02:34:13.680 | for writing amazing books, for being legit and still pushing forward the research science side

02:34:20.400 | of it, and for allowing me and these pothead questions, and also for educating everybody

02:34:29.680 | through your own podcast. Everybody should stop everything and subscribe and listen to every

02:34:36.640 | single episode of Mindscape. So thank you. I've been a huge fan forever. I'm really honored that

02:34:41.120 | you would speak with me in the early days when I was still starting this podcast and meeting the

02:34:45.680 | world. - I appreciate it. Thanks very much

02:34:47.280 | for having me on, now that you're a big deal, still having me on.

02:34:50.000 | - Thank you, Sean. Thanks for listening to this conversation with Sean Carroll. To support this

02:34:56.080 | podcast, please check out our sponsors in the description. And now let me leave you with some

02:35:01.040 | words from Richard Feynman. Study hard what interests you the most in the most undisciplined,

02:35:07.760 | irreverent, and original manner possible. Thank you for listening and hope to see you next time.

02:35:15.040 | [END]

02:35:16.500 | [END]

02:35:17.540 | [END]

02:35:17.620 | [END]

02:35:18.660 | [END]

02:35:18.740 | [BLANK_AUDIO]

Sean Carroll: General Relativity, Quantum Mechanics, Black Holes & Aliens | Lex Fridman Podcast #428

Chapters