– So how then do you make peace with the fact that there are such replicable, and repeatable, and reliable patterns that happen out there?

– If, you know, you have this feeling from your experience of the everyday world that if you see patterns, if you see things that, you know, reliably happen again and again in the same way, there is a reason why, right?

– Yeah.

– It’s not just an accident. And given that kind of attitude, the universe clearly has patterns in it. There has to be a reason why. And so if you’re not gonna say that it’s God doing it, then maybe you’re saying it’s the laws of physics doing it. So I think you have to simultaneously allow yourself to always ask, is there a better explanation? Is there a deeper level? Is there a reason why? And be prepared for the answer to be, no, there is not.

– Hey, everyone. Thanks for joining us. Today’s conversation is gonna focus on the foundations of physics, how the world actually works when you dig as deeply as we have been able to so far. And I’m so pleased that I’m being joined today by someone you all know from his bestselling books. That is Prof. Sean Carroll from Johns Hopkins University.

– Hi, Brian.

– Good to see you.

– Yeah.

– So I figured we would just mix it up and see where this conversation goes.

– There’s a lot, lot to cover. A lot of universe out there.

– Exactly. There is a lot of space. I want to begin with something that I find that many people in our field have a wide variety of perspectives on, which is, well, look, maybe I would say it this way. When I was a kid, by which I mean like when I was 20 or 30-

– Yeah.

– You know, my view back then, which has changed, was that Einstein’s equations were out there governing the universe. Schrodinger’s equation was out there, and like the math to me was reality. I didn’t really draw a distinction between the two. And then over the years, as I’ve thought more about things and become, I don’t know, old or whatever, I now have flipped the other way. I see mathematics as a construct of the human mind that we are pattern seeking creatures, and math is a very good language for encapsulating those patterns, and so that’s what we do. We come up with the language, we come up with the symbols to articulate pattern and that’s what mathematics is, and that’s all that the laws of physics are, a human attempt to encapsulate the patterns that we see out there, nothing more than that. Where do you stand?

– What is the right answer, you’re asking?

– Yeah, that was the right answer.

– Yeah, good. We’re starting at the top here. These are big, difficult issues right at the intersection of natural science and philosophy, right? In fact, I would almost subdivide them. There’s kind of two things going on here. One is the issue of mathematical realism.

– [Brian] Yes.

– Platonism versus something else, sometimes nominalism. We have lots of -isms to describe all these things. I think that many working mathematicians are on the side of your youthful self, right?

– Yes.

– Who thought that mathematical objects real, they’re out there and we’re discovering that.

– And they get mad at me when I say otherwise.

– They do, they do.

– I have to say I’ve had this experience.

– You’ve been tainted by your association with physicists.

– Yeah.

– But, so honestly, like I’m open to that idea, but I don’t see how it would matter. You know, like as physicists, at the end of the day, we want to say like, what difference would it make if it were not true? That’s like a generalization of Karl Popper saying you should be falsifiable, right? Like, I’m not so strict about falsifiability, but I do want to know, does it matter?

– Wouldn’t it matter to your heart of hearts?

– No, it truly does not.

– It really wouldn’t? Because let me give you an example though. So I was at some conference, I think it was in Utah or Arizona. I can’t remember. Some state out there. And I was sitting next to a mathematician, and we just were having this conversation over lunch. And he was at first appalled that I really, truly believed what I was saying. He thought I was just trying to be provocative. And I was like, “No, that’s really how I feel.” And then we continued to talk throughout the conference, and by the end of the conference, I kind of felt that he was having a little bit of an existential crisis. He was like, “I get the argument that this is all made up. I’ve never really felt or believed it. I just sort of dismissed it. But now that I really engage with it, maybe there’s something to what you’re saying.” And then he said, “I wonder what I have been doing, because if the math isn’t out there, I’ve spent my life playing a game, in essence.”

– [Sean] Right.

– “Of trying to solve these various human made puzzles, which ultimately that’s what it would reduce to if math is not somehow anchored to reality.” So it seems like these questions can really matter, at least to some. But I guess you’re saying as a physicist you’re like, “Ah, whatever,” you know.

– Well, I think you’re absolutely right, you and your erstwhile friend.

– Yeah.

– It matters to practice, right? It matters to how we pursue what we’re doing. I’m a big believer and, you know, Richard Feynman has said this, like many other things. He said something lots of people believe, but he said it very nicely that, you know, theories in physics can be absolutely equivalent in their predictions and so forth, but they can be spiritually different.

– Yeah.

– Right? You just capture the same stuff in a different language. And I’m very much of the opinion that we are not finished with physics yet, and presumably we’re not finished with math either, and therefore the direction you might be tempted to move toward to make things better might be heavily influenced by your opinion about, is this real? Am I just making it up? Blah, blah, blah.

– Right.

– And there’s even more, you know, down to earth arguments. You might know, among the many wonderful things in 20th century mathematics, we learned how the old dream from Euclid to Hilbert would be that we would write down the axioms, and they would prove all the theorems and math would be done.

– And then Godel comes along.

– And Kurt Godel said, “No, no, no, that’s not gonna work.”

– Right.

– And one of the ramifications of that was if you have a set of axioms, so you say like, I have a number called zero, and whenever I have a number N, I have a number N plus one, right? You build up the integers this way or the natural numbers. And it turns out that what you’re trying to do is capture the essence of this thing you thought you understood, the integers and you can’t.

– Yeah.

– You literally can’t write down a set of axioms that gives you the integers as you know them and nothing else.

– Yeah.

– And so one argument for mathematical realism is, but I know what the integers are. So whether the axioms capture them or not, I know something about reality there.

– [Brian] Right.

– And it’s not just a matter of sort of describing things that I see in the world. But anyway, but I’m not on that side. I am on the side, I even wrote a paper nudged by your colleague at Columbia, Justin Clarke-Doane, who is a philosopher who does philosophy of math and also other things. Because, so this is a funny story. Justin wrote a book called “Morality and Mathematics,” and it’s about the reality of both morality and mathematics, right? And he starts by saying like, mathematics is the most well-founded thing we have. Morality is the least well-founded thing we have. But in fact, when you look at them carefully, there’s a lot of similarities. And he opens the book with a quote from me, which I thought was very charming, from my book “The Big Picture” where I’m saying, “I am not a moral realist. I don’t think that there are morals out there in the world that really exist that we can test.”

– Yeah, neither do I. Yeah, yeah.

– Right. And then on page two or whatever, Justin says, “But of course, someone like Sean Carroll’s a theoretical physicist, has to be a mathematical realist because he believes in the laws of physics, and they’re expressed in the language of math.” And I told him, “But I’m not a mathematical realist.” So anyway, he nudged me to write this paper that I called “Reality Realism,” which basically was the attitude you just said. What exists is the world, right? What exists is stuff out there, and we invent these languages to talk about it, and mathematics is just such a language.

– Right.

– But to finish the original, original thought, there’s a separate issue, separate distinction that is closely relevant here, which is about not mathematical objects, but the laws of physics.

– Right.

– Are they real as things that bring, that breathe fire into the equations, right?

– Stephen Hawking.

– As Stephen Hawking says.

– Yeah, right.

– So there’s this view called Humeanism, after David Hume that says, again, what we were just saying, no, like the world exists. There’s a bunch of stuff and they’re just doing things, and we invent descriptions of it, and that’s what the laws of physics are. There’s another view, which is called anti-Humeanism, because no one famous wants to attach their name to it, which says, no, no, the world happens, but there are things that you might think of as the laws of physics that generate the world, that they have some power to govern the world and bring it into existence. And again, I see the temptation for that because otherwise, why are there patterns at all?

– Yeah.

– Right? You know, why do you just get random craziness if there’s nothing governing it? But on the other hand, I say like, well, what if there weren’t? What if there was just the universe doing its thing without anything governing it? How would you know? And so at the end of the day, I guess my inclination is always to ask, what is the minimal set of beliefs I need to have to come to an understanding of the world? And I don’t think I need to understand that in addition to the physical world, there are either laws of physics or mathematical objects.

– So how then do you make peace with the fact that there are such replicable, and repeatable, and reliable patterns that happen out there? Now for myself, in the back of my mind, I think I’m a little inconsistent, because I think in the back of my mind, and I do use this language, I speak of the laws of physics. And by that I don’t mean Einstein’s equations or Schrodinger’s equation because I don’t know. They could be provisional and they presumably are provisional. But I do somehow, I think, imagine that there is some uber law that somehow is part of the rock bottom reality. Now, I think all we humans can ever do is approximate that law, you know, get close to that law. Maybe we’ll be lucky and we hit upon that. I don’t know.

– You’re literally a string theorist. You have to think that we could do it.

– Exactly, you know. But I think that is what is behind my intuition. It’s hard for me to imagine having a sense of coherence without that as part of the architecture. Do you similarly have that underneath at all?

– I have the feeling. I get the feeling. I sympathize with the feeling. Like if, you know, you have this feeling from your experience of the everyday world that if you see patterns, if you see things that, you know, reliably happen again and again in the same way, there is a reason why, right?

– Yeah.

– It’s not just an accident. And given that kind of attitude, the universe clearly has patterns in it, there has to be a reason why. And so if you’re not gonna say that it’s God doing it, then maybe you’re saying it’s the laws of physics doing it. The third option is, no, it just is.

– Yeah.

– Right? And so I try to accommodate myself to the belief that it just is, even though I see the temptation to say, no, there’s something causing it to be that way. Because even though I see the temptation, at the end, I don’t see what is gained.

– [Brian] Right.

– I mean, I do think the laws of physics exist as descriptions, not as governing essences.

– But that’s interesting because I think, like we’ve not had this conversation before.

– No.

– But I think we’re in the minority.

– You totally amused me here. I had no idea-

– Yeah, right. Sorry, yeah.

– I was not ready.

– But I think we’re in the minority among our colleagues is my guess.

– Yeah.

– And you know, we’re that, maybe because we both think about foundations and, you know, things that I don’t know. But in terms of the practical utility, if you get really comfortable with the perspective, it just is, do you think that undermines motivation for like pushing beyond? Because at some point you hit a wall, and I guess the question is when do you invoke-

– Right.

– That kind of reasoning?

– Yeah, I think that’s a great question. My attitude toward that is there’s never a point at which you can demand a further explanation. Like whatever you see and believe about the universe, it might be the case that that’s it, like that’s, it’s just like in fact, Nancy Cartwright, philosopher of science has famously put forward this idea the laws of physics aren’t even unified into one big law. It’s just a patchwork, right? There’s the laws that are appropriate here, laws that are appropriate there. I do think, like you, that there’s probably one best underlying set of laws, and I think that those laws exist as descriptions, but that’s just another way of saying that what exists is the world.

– Right.

– And if you want to go further and say, but why does the world exist? Why does the world have the properties it does? I think you have to be happy with the answer that at some point, it’s just like that.

– [Brian] Right.

– So I think you have to simultaneously allow yourself to always ask, is there a better explanation? Is there a deeper level? Is there a reason why? And be prepared for the answer to be, no, there is not.

– Right, right. Now, you know, through the history of science, the history of our species, we have gotten ever better at examining the world, right? In the earliest moments, our brethren looked upward, and they used their eyes to see what they could, their ears to hear what they could and so forth, and so they had a limited data set to try to explain and gain a sense of what happens in reality. As we march through the history of our species, we get better, we realize that the light we can see with our eyes is a small part of this gigantic electromagnetic spectrum. There’s so much more out there. So do you spend any time worrying about, and the answer may be no because there’s not much we can say about it, but do you spend any time worrying about the possibility that the patterns that we are aware of are the patterns that we can be sensitive to, either through our own senses or the equipment that we build. Maybe that’s like a slice through this gigantic reality that we’re not at least as yet privy to, and maybe therefore the patterns that we’re seeing are like misleading. You know, maybe there’s a whole rich reality out there that just violates what we would have thought based upon the things that we can directly sense.

– So I think that’s 100% a viable possibility, and I spend zero time worrying about it.

– Right.

– I mean, I would absolutely buy an argument that says that’s already happened with the transition from classical mechanics to quantum mechanics.

– Yes.

– Quantum mechanics, as I think about it, is very much a story of stuff that is entirely alien to the everyday world in some very real sense, and there’s this complicated story to be told about how to connect it to the world, and that story implies the existence of many, many things we can’t see, and that we are only experiencing a tiny sliver of reality. But we were drawn to that, we were forced to that by trying to explain the world that we do live in.

– Yeah.

– And we do see. And so I think that it’s completely plausible that we’re driven to some picture because it’s the best explanation or account of the tiny bit we do see, even if what we do see is a tiny bit. If there’s another level where there’s a whole bigger reality that just has no impact on what we see, then I’m less interested.

– Yeah, for sure.

– Like, yeah, okay, maybe, but I mean, then it’s a simulation argument or something like that.

– Right.

– I remember there was a workshop at the Santa Fe Institute, where I’m a part-time faculty member, and it was called Limits of Explanation. So you know, at what points do we just reach, you know, our human inability to go any further in explaining the world? And we were chatting back and forth and, you know, having, and at some point it dawned on me, like, I’m like, “Wait a minute. Do you people think that we’re close to the limits of explanation?” And they were all like, “Yeah, we might be very, very close. We might be running out.” I don’t think we’re anywhere close to the limits.

– And what gives you that confidence? I mean, I think I share that general perspective, you know. I mean, you know, the example that I like to use was actually in a little section of a Nova program that I did years and years ago, where I was at a blackboard lecturing general relativity to some student who was clearly not getting it. I was getting frustrated, and the camera pans and it’s a dog. You know, and the point which the audience seemed to miss, unfortunately. I wasn’t trying to say that they’re like the dog. I was trying to say dogs are smart, but they have intellectual limits, we think, on what they can understand. Although every time I say that I wonder if the dogs are barking at us humans-

– They’re smarter than us.

– They don’t know. They haven’t gotten the unified theory yet. But putting that to the side, you know, they have limits, so we have limits. But we’ve never seemed to hit the wall, and we’re always able to go forward. So I feel that like you, I suspect, very optimistic that we have a fine future ahead of us as, you know, a species of explorers. But yeah, I mean, there’s every reason to believe that there are limits to what this brain that evolved for a very different purpose. It didn’t evolve to understand physics. It evolved so our forebears could survive in the African savanna. So why would it be fine tuned for these kinds of questions?

– Well, I do think and there, I don’t know how common this view that I have is. I do think that there’s been a phase transition somewhere in between the dogs and us, where we are somehow Turing complete. That is to say in the language of computation theory, Alan Turing.

– [Brian] Yep.

– And I’m only using this in a sort of vague sense, not a very technical sense. But he had this idea that there was a set of functions that could be computable by a computer, and they’re called the Turing complete functions. And some computers just don’t have the ability to calculate all of them, but once you have a big enough or powerful enough computer, there’s no limits. You can just calculate all of them. And I do think that human beings sort of cross some threshold of an ability to abstract where I don’t see any evidence, other than a kind of humility, that there’s another threshold out there that we haven’t yet crossed-

– [Brian] Yeah, right.

– That would allow us to understand further things. And again, maybe it’s there, but the data is not forcing me to worry about that one.

– Yeah, no, it kind of brings to mind a thought that I have, which I’ve not shared much about because I can’t make it precise. But just among friends, you know.

– Yeah, no one will know.

– No, it feels to me, like maybe this is related to that phase transition, but it feels to me that a critical moment in our intellectual development as a species was our ability, at some point, to invent metaphors. It feels to me that metaphor is the moment when you say, you represent something real by a word that no longer is directly tied to its reality. And so once you can symbolically represent the world, now you can write down an X for a position, or an F for a force, just to be flatfooted in the metaphorical representation of real things in the external world. And to me, that is where mathematics ultimately comes from. It’s a metaphorical description of things that are actually happening out there in a poetic language that still allows us to calculate and make mathematical predictions, because we can then go backwards or forwards in this metaphorical description of things. That to me is a moment. Now, I’ve discussed this with a few people, and they’re like, “Eh, I don’t know, you’re making too much of metaphor.” But it feels to me that that’s kind of what we do.

– I don’t know. Yeah, so actually, I’m more or less completely on board with what you just said. I also cannot put it in precise language, but I do think that there’s something in the combination of metaphor, symbolism, abstraction, imagination, counterfactual reasoning, logic, like all of these things, language itself, right, that gives us a way to think about not just the world, but possible worlds in a very, very powerful, abstract way.

– Yeah.

– You know, there’s a famous story by Borges, “The Library of Babel,” right?

– Babel, yeah, I know it well.

– Where he says-

– I teach it to my students.

– Excellent. Good. So let me ask if you tell them the following thing.

– Probably not, but let’s hear it.

– Well, for those who don’t know, the story is, you know, you just have a library with a set of books of some fixed length. Do you remember what the length is supposed to be?

– 440 pages.

– There you go.

– I probably got that wrong.

– Okay.

– Someone will correct it.

– And basically you just fill up the books with, someone filled up the books with every possible sequence of symbols, right? So there’s many, many books. If you can calculate the number of books, it’s quite large, but you have a finite alphabet, a finite length of the books. There’s a finite number.

– Yeah.

– And somewhere in the library is every book ever written or ever will be written. And you might say, “Well, what if my book is 500 pages long?” Then it’s just two books. You say one book. But okay, and there’s lots of fun things to say about that. But then if you step back and think about it, we have made the Library of Babel, but rather than a collection of books, it’s called the alphabet. Every book that could ever be written is implicit in the alphabet. All you have to do is just rearrange the letters.

– Rearrange the terms, yep, yep.

– Right? But what hopefully that drives home is that rearranging the letters is the hard part. The existence of the symbolism is a tool-

– Or finding the card catalog is the hard part.

– The card catalog is the size of the library, sadly, right?

– Yeah, right, right, exactly, yeah.

– So, yeah. So, like, it’s combinatorics that does it, ultimately. The fact that when you have symbols that can take different values, and you arrange them in a string, the number of possible strings just blows up to some huge number very, very quickly. Much more quickly than things like the number of atoms in the universe.

– Yeah.

– Right? So we’ll never explore the space of Borges’ library, Borges’ library. What we’ll do is carve paths through it, and that’s where we started with, you know, thinking about math or physics in different ways can suggest different ways to explore this vast space of possibilities.

– Yeah. Now, within that library, of course, there is, there are deep truths in some of these books, and the hard part is finding which books have the deep truth that’s of interest to whatever issue you might be facing. We have of course encountered a variety of problems in theoretical physics, for which it’s not that we haven’t yet made progress. It can be hard to imagine how we would really make progress. You know, the big one perhaps is why is there something rather than nothing, right?

– Exactly.

– Leibniz asked this question a long time ago. And within the paradigm of modern science, it seems that we don’t quite have the tools to even address that question, because you look at any theory of modern physics, you got to start with something, right? You got to start with, you know, some ingredients, some symbols of space or time where we’re getting a little bit beyond having to, you know. But when it comes to questions like that, that of course is the place where there’s some who’ll say, “Well, that’s like where God comes into the story.” And frankly, if you’re gonna limit God in that way to stand outside the place where physics operates, I’m like, okay, you know, that doesn’t affect things beyond a sort of a deistic perspective. But I’m not drawn to that way of thinking about things because I just feel like, as we were saying, we can make progress, and we can go further back, and we can shrink the gaps in our understanding and so forth. Are there questions where you would imagine that that, you know, a deistic approach would have any purchase on you? Or is that just no way. That is just a cheap, lazy way out, and I’m not gonna go there ever?

– It almost has none, to be honest. I’m not tempted by that. Like I get the temptation to think the laws of physics are real because the universal bases patterns. I am completely untempted by invoking some deistic God to provide an explanation for things like why the universe exists rather than doesn’t, precisely because then you have to say why does that God exist rather than doesn’t? And believe me, I went to Catholic school.

– Oh, is that right?

– The Pope and I are, you know, we went to the same undergraduate institution.

– To what grade? To what grade?

– College, college.

– Really?

– Yeah.

– Wow.

– Yeah, yeah, Villanova.

– Oh, amazing.

– Me and the Pope, we’re very close.

– No, so then what was that like? Were you practice-

– Well, my point is

– Were you practicing?

– I was totally non-Catholic. Villanova let you be non-Catholic, as long as you took a lot of religion courses, which I had to do, so.

– But that was potentially interesting or was it really doctrinal-

– Super interesting. Or was it more-

– No, super interesting. Look, my wife Jennifer, who is an ex-evangelical-

– Yeah.

– You know, went to a deeply religious evangelical college, and she says the most rational people there were the religious studies professors, because they cared about the subject matter, rather than just using it as a symbol for something else.

– [Brian] Right, right.

– So yeah, no, my religious studies and philosophy professors at Villanova had thought very carefully about all these things. Many stories there to tell. But the point I’m trying to get at is I know very, very well the counter arguments to when I say, “If you’re explaining the universe as a contingent fact by invoking God as a necessary being, why can’t I just say God is contingent?” And they do have counter arguments, and I think that they’re entirely unpersuasive. And I think, again, it doesn’t get me anywhere. It doesn’t lead anywhere interesting. I do think that the question of why there’s something rather than nothing is the leading candidate for a question to whom the answer is, “It just is, you just got to accept it.” And maybe there’s a better answer, but it’s not gonna be a satisfying answer. It’s gonna be an answer of the form, “You’re not allowed to ask that question because there’s no such thing as non-existence that we could have had as another option.”

– Right, because there can be questions that satisfy the rules of English grammar.

– Of grammar.

– But yet they may not be as sensible as they illusorily seem to be, because language can be misleading and makes you-

– That’s exactly it.

– Makes things of that sort. So, you know, when you think back to Villanova, you know, no doubt like I, you’ve had various debates and conversations over the decades about these issues. Is there any part of you that has even the slightest envy for folks who really do believe? Because it gives a, you know, ready-made sense of meaning or purpose, or that death is not the end or, you know, or does you just like, I just don’t care about that?

– No. No part of me is envious for two reasons. One reason is that I’ve seen people who are religious in situations of tragedy or things going badly. They don’t handle it any better than atheists or agnostic do, as an empirical matter. Like they’re just as upset by death and whatever. But more importantly-

– But long term, have you not seen that someone’s faith, you know, however much… Let’s not even judge whether it’s objectively true, but the subjective belief can at least, in some cases, help someone get through a traumatic period by virtue of seeing it part of a greater plan.

– But that’s the thing, it absolutely can.

– Yeah.

– And it can absolutely torture you for no good reason.

– Sure. Yeah, yeah, yeah.

– So the problem is exactly for me like just moral objectivity or moral realism. You know, wouldn’t you feel better if you thought that your moral rules that you were following were out there and objectively true in the world? Well, what are the right ones, right?

– Yeah, yeah, yeah.

– Like what are the beliefs that you have, if the beliefs you have are not actually correct beliefs, then there’s no guarantee that you’re gonna have beliefs that will help you or help the world. Now I completely, I’m not one of these atheists who thinks that religion is absolutely bad for everyone under every circumstance.

– Should be wiped off the face of the earth, yeah.

– Well, I do, I mean, part of me thinks it should be wiped off-

– Really?

– In the sense that I think that it is wrong, and I think that I would be, the world would be better if more people were correct in their beliefs.

– But-

– I don’t want to go around wiping it out, but I want everyone to have true beliefs.

– But when you say wrong-

– Yeah.

– Presumably it’s on the very specific scale of describing an objective reality. But if you have a somewhat more nuanced sense of truth, there’s like objective truth that everybody, and then there are subjective truths that we just hold inside. I guess my view has been if somebody holds within their worldview, but the inner way in which they try to make sense and find coherence in reality. If they’re holding to one or another religious system and if it works for them, I’m saying that’s great for them. And moreover, it has always felt to me that, like, I’m not religious and I don’t believe that there is any objective truth to these kinds of things. Nevertheless, historically and evolutionarily, it feels to me that religion can be viewed as a remarkable invention of the human species, right? We face death. What do we do about that? Our ancestors on the African savanna trying to make sense of a world in which the people they care about are here one moment and then gone the next, to invent this idea that there is this other realm where they may join together in some future. To me, that’s a interesting and powerful story, if you take it as a symbolic way of engaging with the world, and not try to use it to explain why the earth goes around the sun or things of that sort. But does that not help you give it a kind of existence that is worthwhile?

– So yeah, like, I have at least three things I need to say.

– Yeah.

– I’m gonna try to keep in my brain before I run out. One is of course as an empirical matter, I know a lot of religious people who are better people than certain atheists.

– Right.

– And vice versa.

– Sure.

– Right? So when I say I would have, I’m in favor of, you know, eliminating religion from the world, in practice, sure. There’s plenty of people I know who are religious. We’re friends. We joke about it, you know? I’m not trying to change their minds all the time. I’m not berating them or anything like that. As an empirical, real world, human beings are complicated matter.

– [Brian] Yeah.

– Second thing is if, to the extent that it’s a coherent thought experiment, if I lived a thousand years ago, I’d definitely be religious.

– Wow.

– Like, it was obviously the right, best thing on the market at the time.

– Newton was, right?

– Yeah, Newton.

– Galileo. You know what I’m saying?

– Yes, that’s right. Galileo, harder to say, but yeah. Newton definitely-

– No, I think with Galileo, it’s quite clear. He wasn’t trying to go against the church, per se. He was trying to tell the church, let us as human beings use our God given gifts of analysis and perception to gain a deeper understanding into God’s creation. That was really-

– Those are the words he said, yes.

– You know, but somewhere in there is I think where his perspective was.

– It’s possible. It’s possible. I just think like as a matter of personality, like Newton, I can clearly see-

– Yeah, for sure.

– Was of like. Galileo, I think there’s a large phase space, as it were, where he was just telling people what they wanted to hear when he said those words.

– Maybe so.

– Maybe not also, but I think it’s possible. But the third thing I want to say is, you know, I do, and this is the romantic side of me coming out. I like the truth.

– Yeah.

– I’m in favor of it. I think that in general, like I don’t think you should always say the truth all the time. Like when someone says, “How does my new haircut look?” you’re allowed to tell a little white lie. If there’s some political, social project that you’re engaged in, and not being completely truthful leads to a greater good, I’m not, you know, too strict about that either. But as a general rule, I mean, you’re saying, isn’t it, you know, better for certain people to have these religious beliefs if it coheres inside themselves? I absolutely believe there’s a way that that person could have a coherent set of beliefs that did not involve religious commitments that would still remain, still having remaining them, still have them remain being a good person.

– Yeah, yeah. Like I don’t think that behavior is tied necessarily to holding some kind of religious beliefs. I totally agree with that. But I’ve certainly had conversations with people for whom the absence of belief would be such a radical tearing away of the fiber of what makes their life feel meaningful. Again, it’s not for me and clearly not for you, but I respect that that is the makeup of their particular way of living in a absurd universe is sort of how I look at it.

– But do you think that anyone says to themselves, “This isn’t true, but I’m gonna believe it because it makes me feel better?”

– That’s interesting. I don’t know. I haven’t sort of framed the question that directly. But I guess if I-

– Or is it a little patronizing to say, well-

– It could be.

– You believe those illusions. It makes you feel-

– Yeah, yeah, yeah. It certainly can be, no doubt about that. But let me just use myself as an example, actually. I sat here before and said I don’t believe, you know, that’s not, but if I’m being fully honest, not this is a therapy session here, but there is part of me at moments that buys into something that I know is not true because it feels better at that moment. I mean, there definitely have been times when I’ve prayed. I mean, I didn’t go to temple-

– Yep.

– To sit and actually be within the institutional structure. But yeah, there’ve been moments when I’ve actually said to myself, “I don’t know if it’s real. I don’t think it is. But if you’re out there, Mom, you know, protect, or if Dad,” you know what I’m saying?

– Yeah.

– So, in some sense, that’s deeply incoherent.

– Right.

– It makes no sense from any kind of objective standpoint. And yet as a logical and generally coherent person, I have found myself doing that. So I think I was doing exactly what you’re asking.

– Well, I don’t have that temptation really at all.

– So you never, you never do that?

– No, I really never do.

– Okay.

– I mean, the more general temptation to allow-

– So Dr. Carroll, what’s wrong with me?

– Well, one picks and chooses, there is always a temptation to believe things that in your more rational moments you would see are not true. For me, that does not happen in praying. It happens while playing poker. And I’m saying, I do not believe that person has a full house.

– Right.

– And in retrospect, I really should have believed that they did, right? Made me feel good in the moment.

– Right.

– But, yeah, so that’s human nature, and I, again, I want to know what the total set of implications for those kinds of beliefs are. They might be entirely harmless or even entirely beneficial.

– [Brian] Yeah.

– Again, I don’t think that there’s a self-consistent inner life which says, “I know these are wrong, but I’m gonna believe them,” right? People say like, “I know other people don’t believe it, but I believe it.” Or they say “I’m not sure whether it’s right or wrong, but I’m gonna believe it.” But I don’t think that people generally say, “I know that the evidence and the canons of rationality say that I shouldn’t believe X, but it makes me feel good, so I will believe it.” I think the people who believe these things believe them and think that they’re right.

– Right. But I think the journey to that, and again, this would be an interesting question. I would just like to know, I mean, you know, someone like Bill Phillips.

– Yeah. Yeah.

– You know? You know, I’ve had some conversation, not in a long, long time, and, you know, here’s someone won the Nobel Prize in physics, so has a grasp of the basic laws and how they work, and yet is deeply religious.

– Yep.

– And there’s no inkling of these things being in conflict with each other, because I presume he’s able to apply each in a domain that is relevant for the precepts and conclusions of that structure.

– Right.

– I don’t think he uses any of his religious ideas when he’s doing his physics experiments, and I don’t think he used any physics experiments to shed light on the reality of his religious belief. So somewhere in there is this balance.

– Yeah, no, actually that, I’m glad you brought up that specific example because that I have stronger objections to, in fact. I mean, this is sort of a version of Stephen Jay Gould’s famous-

– Magisteria, yeah.

– Non-overlapping magisteria, which I think is just not right. I mean, that’s just pulling the wool over your own eyes intentionally. If I thought that God existed in some way that was vaguely related to traditional monotheism, like, you know, a personalized God-

– Right.

– Who loved the world, in some sense, and was all powerful-

– Created us in their image, yeah.

– Et cetera. How in the world could that not affect the way that I think about physics?

– Right.

– You know, my physics is trying to understand the world at the deepest level, and to just say like, “Oh, that has nothing to do with my belief in how the world works at the deepest level.”

– Yeah.

– Seems a little incoherent to me. I saw that I’m more sympathetic to people like Don Page, our mutual friend-

– Yeah, sure, yeah, yeah.

– Who is a cosmologist and a born again Christian, who is very happy to mix his religious beliefs with his cosmology, because of course they’re asking the deep questions. Why wouldn’t they be related?

– Yeah, right, right. So from that point of view, it’s somehow more truthful to bring them all together.

– I mean, I do think that we human beings, and I’m sure that I’m included in this, aren’t great at making sure that our various sets of beliefs are mutually consistent.

– Yeah, right. I think one of the wonderful qualities of being a human being is our capacity to hold mutually conflicting ideas, and somehow allow them to intermingle, and develop, and change our mind over time and so forth. You know, another deeply related issue to this, which I think we disagree on, maybe we don’t, but I’ve heard you say things that make me think that maybe we disagree, is the question of free will. So you know, my view, I think it’s kind of similar to what many in our field would say, which is, you know, I look at myself as ultimately a collection of particles guided by physical law. And I don’t have any, I, I don’t even know what I means at that level, I should say. But in some vague notion of personal identity called I, I don’t seem to have the ability to intercede in the lawful unfolding, whether it’s classical or quantum mechanical, hardly matters actually for this question. And so I don’t seem to see a place for me to intercede in the lawful unfolding in a way that my intuition suggests that I do. Now, I, of course, recognize that there are different levels of explanation in the world, and free will is like way up high. You know, it’s at the level of trying to understand the aggregate world, and one could say, “Well, it’s just a word that has no instantiation, no means of applying it right down there at the fundamental particles.” And yet it does feel to me that the levels need to cohere in some kind of logical respect for what each allows. And if the laws of physics don’t allow intersection in the fundamental unfolding of the particles, when I lift up my hand, it’s just the motion of particles and therefore I don’t have the control that I would’ve though based on my intuition. So that’s the lack of freedom of the will. Where do you stand on that?

– I think you’re so close. You’re gonna get there. We’re gonna get to there. No, I think you have to respect the levels. I don’t think that you can say, like everything you said basically I agree with. We’re made of particles, particles obey laws of physics. But then you want to say, “I can’t intervene to change what the particles are doing.” That’s just a grammatical mistake.

– Because the I is only at the aggregate level. Yeah, for sure.

– Yeah, there’s a higher emergent level and it plays by its rules, and that relationship between the levels is one of consistency.

– Sure.

– Right? So I, as a aggregate agent at the level of the microscopic world, of our, you know, folk physics or whatever you want to call it, I don’t know what I’m gonna do next.

– Yeah, yeah, totally, totally.

– And indeed, when I try to do things in the world, I do something called rational decision making, or at least I try to approximate that.

– Right.

– And everyone who denies the existence of free will spends all of their time trying to make people make choices about things. Like everyone operationally acts as if it’s there.

– Totally agree.

– I just think that’s it.

– But you do agree then that if I were to analyze you at the level of your particles, and I sort of throw away momentarily the higher level aggregate structures called Sean Carroll, the brain of Sean Carroll and all that sort of stuff, that I would have no barrier to understanding the motion of those particles totally with the absence of anything like Sean Carroll’s decisions or choices interceding in the lawful unfolding of the particles.

– [Sean] Right.

– So if I saw a collection of particles that on the aggregate we call your left arm, if I saw that aggregate of particles go up, I would not lack for any explanation if I was solely at the level of the fundamental constituents.

– That’s completely true.

– Okay, so that’s, yeah, so-

– And what you’re saying is-

– It almost comes down to a language thing.

– Well, it’s not, I think it’s a little bit more practical than that, but it’s also… Oh, sorry, it’s not just practical. I think it’s a little bit more oomphy than that.

– Okay.

– You know, this is Daniel Dennett’s idea of real patterns in the world, and this is the crucial feature of emergence, and this is more is different, a la Philip Anderson, our physics colleague.

– Yep.

– And the point of all these buzz phrases is I don’t need to know about the lower levels to have a good theory of the higher levels.

– [Brian] I agree. Yep.

– And indeed, I might have various, there’s multiple realizability. Maybe there’s different kinds of lower level theories that give rise to the same higher level phenomenon.

– [Brian] Right.

– And in practice, you know, it’s very tempting to say, well, if I knew the position and velocity of every particle, if I were Laplace’s demon, in other words, then I could do this.

– Yep.

– Yes, you do not know that.

– Sure.

– And you are not Laplace’s demon and you never will be.

– As a practical matter. Yes, yes. I agree.

– Even in principle, you will never be able to do that.

– Right.

– And so what?

– Right, so, so I totally agree with that. The one thing that feels like it still has a significant pull on me is that, and you’ll say it’s an incoherent set of assumptions, which I respect that response, but were you a so-called philosophical zombie?

– Yeah.

– You know, were you you in the absence of the inner world, which I’m just hypothesizing that you have inside of your head right now, and you’re doing the same for me, which I appreciate. But without that inner world, there would be no change in what you do.

– Yeah.

– And so in the absence of the thing that we call the inner world of making decisions and choices and so forth, nothing would change in your behavior in that case.

– Right. So I think that my solution to that is simply the idea of a philosophical zombie-

– It’s incoherent.

– Is not conceivable.

– Yeah, good. Yeah.

– David Chalmers wants to say like, maybe it’s not possible in the world.

– Yeah, yeah.

– But I can imagine something that behaves exactly the same way as a human being or an agent-

– Right.

– But lacks inner experience.

– [Brian] Yeah.

– I would say that as a physicalist, as someone who thinks that the world is basically the physical stuff in the world, you can’t conceive of that.

– Right.

– Because if you had literally every particle in your brain doing the same thing-

– It would have to have-

– It would have the same inner experience.

– The same experience. Yeah, yeah, yeah. No, and I agree actually with that. I find it a useful though experiment to get there, but as I said, the notion that it may be an incoherent set of assumptions is certainly a reasonable response to that.

– And honestly, what I try to do is to say, when people want to talk about free will, I always ask, “Are we allowed to talk about this without using the phrase free will?”

– Right.

– Because suddenly, all the disagreements go away.

– I agree. That’s what I’m saying is in some sense, only the language-

– It’s only the phrase. Yeah.

– That’s right. It can be a language thing. When you think about consciousness then, because this obviously takes us right there, I presume that you view consciousness as this natural outgrowth of a certain set of processes, whether we want to use the language of information theory, or physical instantiation of the laws of physics acting itself out in a particular gloppy, gray structure crenelated inside of our, you know, this bone cage on top of our shoulders. You presumably just see that as something that happens when you’ve got that level of complexity going on.

– Yes.

– And is that enough for you? I mean-

– Yeah, I mean, I think I would just remove the word “just” from your statement. I mean, there’s a lot going on, right?

– Yeah, yeah, yeah, yeah.

– 86 billion neurons bouncing together in your brain. The connectome is very complicated. We have no conception of it. You know, the world often has these properties that there are ways of talking about the collective behavior of a whole bunch of little things at this aggregate level in ways that are surprising and emergent and super duper useful. You know, the law of supply and demand is nowhere to be found in the standard model of particle physics.

– Yep.

– And it more or less obviously emerges in ways that we can understand. So I have enormous respect for the real neuroscientists who are trying to find out how what we call consciousness relates to what happens in the brain.

– Yeah, the neuro correlates of the things that we experience.

– And yet, even though we are not close to that yet, I’m convinced we will get there someday.

– Yeah. No, I am too.

– Yeah.

– And what about instantiating consciousness outside of this biological film?

– Ah, so that’s a lot more fun because now we have to confront it, right, like in a way that five years ago we didn’t. We’re building these things that certainly pass the Turing test, right? Our friend Alan Turing, who we referenced earlier, had this idea, very, you know, very sort of engineering kind of idea that what it would mean to be thinking is to be able to fool somebody else that you’re thinking, right? And somehow people have sort of moved that into consciousness and said like to be conscious, you have to be able to fool someone into thinking you’re a conscious being through a, you know, talking over a computer or whatever. And we have LLMs.

– Right.

– I was talking to one earlier today that can certainly mimic human conversation very, very well. And there are people working for the AI companies who are either already on board or really close to believing that these things are basically conscious.

– Yeah.

– And I entirely strongly, disagree with that.

– Yeah, me too.

– And I think that, and I’ve had people, you know, I have a podcast-

– Yeah.

– “Mindscape” where they-

– No doubt you’ve spoken to people in the field who-

– I have.

– Yeah.

– But even better, my listeners give me a hard time. They’re like, “Why should I listen to you about whether LLMs are conscious, rather than the people who are programming the computers, right? Like you’re not an expert. You always tell me not to listen to the pundits, listen to the experts.” And the thing is that those people are not always experts in intelligence or consciousness.

– [Brian] Of course. Yeah.

– And so I don’t think that it’s necessarily that the argument goes through. There are people like Ned Block here in New York at NYU, and Anil Seth at Sussex who have, in ways that are just like targeted to make me happy as someone who cares about entropy and the arrow of time, really pushed the idea that part of what is important in our conscious experience is the process of the passage of time at a microscopic level, you know, below the surface, right? Like LLMs, large language models, don’t get bored. They don’t experience the passage of time. You can just turn one off, turn it on. It just pops up just as well, right? And we have all this biology going on, all these little mitochondria going on, making ATP and things like that. And of course we want to abstract and, you know, say, well, okay, but at the higher emergent level, does it really matter that all these biological processes are going on? And people like Block and Seth are saying, “Actually, yes, that’s actually really, really important.” And I think they are tempted by going all the way to saying you need biology to be conscious.

– [Brian] Right.

– And I’m not going that far, but I might be persuaded by the idea that you need an inner process that increases entropy to be conscious.

– But can you just simulate that? I mean, do you need to actually-

– Maybe, yeah.

– Have it in this form or-

– So I suspect that you can simulate it.

– Yeah.

– I also strongly suspect that simulating it is gonna be much harder than people think, because we’re really, we’ve optimized to give the surface experience of talking to an agent, right? We’ve not optimized do all the stuff underneath that gives rise to that inner life, right?

– Right.

– There was a recent, I don’t know if you saw this fun study from Anthropic, one of the companies.

– I did, well, I think I know what you’re referring to.

– That, you know, identified emotion vectors.

– No, I don’t know this one, yeah.

– Yeah, they can actually like do a projection onto certain states of their LLM, and characterize it as angry or sad or whatever.

– I see, by like the frequency of certain word use and things like that?

– It’s literally like the weights of, you know, the state of the LLM at one moment of time, and that affects its responses. It’s angry, it responds in certain ways, okay? So there’s a predictive, useful characterization of what is going on in the AI. Does that mean that they are angry? I think no.

– Right.

– I don’t think it means that. I think that, you know, again, we’re really being very, very clever about simulating very subtle things, and we’re always super quick to anthropomorphize, right?

– [Brian] Sure.

– Like since we’ve never met things that acted conscious that weren’t.

– [Brian] Yeah.

– Now we meet these things and we’re gonna attribute that, but I don’t buy it.

– And evolutionarily, it makes sense that we anthropomorphize because in the ancestral world, if you fail to-

– True. Right.

– You know, to assign agency to something in the environment, it could kill you, right?

– Yeah.

– So it’s better over than under ascribe agency.

– It goes back to the question about, you know, are we smart or rational enough to understand everything in the world? And I sort of put forward the idea that maybe we are, but man, in practice, we’re very loaded with biases, and intuitions, and assumptions that don’t bear close scrutiny. So that’s why quantum mechanics is hard for us.

– Yeah, yeah totally.

– And that’s why consciousness is hard for us.

– But can that go the reverse too, right? So we have an intuition that consciousness is pretty special, right? And the thing that happens inside our head we feel has a certain kind of wondrous mystery to it. Could it be that we are just giving ourselves an aggrandizement that we don’t deserve? And in the end, it won’t be so hard for some artificial system, you code it correctly and it really is thinking and feeling and it’s like, hey, this just happens a dime a dozen in sufficiently complex systems.

– Yeah, so again, with my podcast listeners, we have a rule that if any question they ask me begins with the phrase, “Is it possible that?”

– The answer is yes.

– The answer’s always yes. So is it, you know, conceivable that it’ll be easier than we think? Yes, like look, I do not think LLMs right now are conscious, but also they’re way better at doing what they do than I would’ve predicted five years ago. Like I was out of, you know, that’s one place I would’ve been very, very wrong.

– [Brian] Right.

– So therefore, yes, I might be completely wrong about this. I do think, so, you know, in other words, I’m pretty strongly of the opinion that LLMs as they currently work are not conscious.

– Yeah, yeah.

– But what you’re asking is the obvious and important follow-up question, so how hard would it be to modify them so they were? And that I truly don’t know if it’s like really, really hard or really, really easy.

– Sure, sure. So, is it possible that there’s a logically coherent explanation for everything that’s internally inconsistent? I want you to say yes, because you said if it begins with.

– That only applies to physics questions.

– Good, okay. There we go.

– Not to logic questions. So the answer is no.

– Yeah, yeah. So when we try to actually, you know, make sense of the world as it’s been presented to us, we are taken to quantum mechanics.

– [Sean] Yeah.

– And you know, there have been arguments about quantum mechanics, and what it tells us about the true nature reality, go all the way back. I mean, it’s a real fascinating literature that you’ve probably immersed yourself in, with Einstein, and Bohr, and Heisenberg, and Schrodinger all sort of going at it from different points of view. Today, those arguments still persist. Some of us, like you, believe, I think, that it’s kind of clear where the reality is. So, you know, obviously you’re a many worlder. I don’t know if that’s, is that a-

– [Sean] That’s fine.

– An acronym that we can use. So give us, you know, obviously, you know, I and many people watch this are at least tangentially clear on the history that, you know, Hugh Everett, 1957, looks at the equations of quantum mechanics and says, “Just take the equation seriously.” And right there, term by term, you see things that call out, cry out to be interpreted as different realities.

– Right. Right.

– Take us through that and further and convince me.

– Yeah-

– Because I’m not convinced.

– Okay. I’ll do it. We can do it.

– Yeah, yeah.

– It takes like two minutes.

– Yeah.

– But first I want to say like as much as I am impressed and a fan of the Everett interpretation of quantum mechanics, AKA Many-Worlds, even more strongly, I want more physicists to care about the foundations of quantum mechanics on that.

– Yeah, yeah, I’m with you on that. Yeah, yeah, I agree with you.

– Whatever their field, whatever their, you know, particular favorite is.

– And along those lines, can I just sort of say, and again, I’m not sure you agree, but I would like people to stop talking about interpretations of quantum mechanics.

– Also agree with that, yes.

– I would like them to talk about different theories that in many cases yield the same predictions, but the structure of a theory matters.

– It does. Yes.

– To what it says about reality.

– Right.

– So it’s not just predictions and all of the rest is interpretation.

– Yeah, the word interpretation’s been holding us back for exactly this reason.

– I agree. I agree.

– So that’s why I think it’s good that people are increasingly using the phrase foundations of quantum mechanics.

– Yeah, right, right, yeah.

– So, okay, as you said, you know, poor Hugh Everett was a grad student in the 1950s with John Wheeler as his PhD advisor. And Wheeler gave him the following thesis topic: quantize gravity. So you’ve tried to do this, right?

– I have spent a long time trying to do that.

– And it was back in the day, and so we didn’t have string theory or anything like that, but Everett, you know, was conscientious and he said like, “What do you even mean quantize the whole universe at once?” Because in the Copenhagen version of quantum mechanics, and this is literally, it’s kind of hilarious, this whole debate has reared its head again very vividly in modern quantum gravity and cosmology.

– [Brian] Yeah.

– The fact that Copenhagen imagines that there is an observer external-

– To the system.

– To the quantum system. And Everett said in the universe, there’s no observer, like he wasn’t religious, I guess. So he says, “Okay, what do you do?” And like you indicated, follow your nose. You have an equation, the Schrodinger equation, which tells you how the quantum state of the universe evolves with time. And look, the short version of it is if you, the really crucial step is do you think that the wave function, this mathematical object that represents the quantum system-

– Yeah.

– So the wave function of an electron, the wave function of a solid or whatever, do you think it represents reality, or do you think it’s just a way of calculating things?

– Right.

– So the Copenhagen people say, “It’s just a way of calculating things,” and that is a get out of reality free card in a lot of ways. And Everett says, “No, let’s just, it’s the simplest thing. Like we need it, this wave function. It obeys an equation, just like the electromagnetic field does. Like what if it’s just reality? What do we do?” And then very directly you say, look, this wave function represents superpositions of different possible answers to questions, like an answer to the question, “Where is the electron?” So the wave function says, “Well, it might be there, it might be there, it might be there, it might be there,” with different weights, different amplitudes. It give you different probabilities for measuring it. And if that’s reality, then when you measure the electron, it’s a immediate consequence of the equation that there is a big wave function that has you in it and the electron, and part of the wave function says the electron was here and I measured it there, and part of it was the electron’s there and I measured it over there, et cetera. And the real crucial philosophical step was because, well, I should say the reason why that’s not obviously true or self-evidently true is that if you stare at that equation and what it’s telling, you might say, but okay, then I, when I make a quantum measurement, should feel like I am in a superposition of all these different possibilities, and no experimenter has ever felt that way. You see definite measurement outcomes.

– [Brian] Yeah.

– So the important philosophical move that Everett made was to say you have misidentified yourself in the quantum state of the universe. It’s not that you are the superposition of all these measurements. Each measurement is a different world. And there’s a version of you that saw the electron there, a version of you that saw the electron there, et cetera. Now, I will say two things at once. One is that at the level of the postulates of the theory, the axiomatization of the formalism, et cetera, it is literally impossible to conceive of a simpler version of quantum mechanics.

– [Brian] I agree. Yeah.

– It is the most direct, most austere, most pure version of quantum mechanics. At the level of matching it onto our experience, it is the hardest.

– [Brian] Yeah.

– And I think it’s completely okay as a sort of methodological principle to be conservative and say, “Look, if your theory is so different than what I observe in the world, I should be suspicious of it.” And I think that’s fine. But then you try to make alternative theories, which people have tried to do, and oh my God, they all look so bad that, you know, that’s the best evidence that Everett is right, that all the other theories look so bad in my mind. But it actually, I don’t know if this is all part of your master plan, I’m sure, but it circles back to what we were saying before, and I think this is the best single argument for Everett, that taking Everett seriously, taking seriously the idea that the fundamental theory of the world is a quantum state that just obeys the Schrodinger equation, versus the idea that there are observers, and they make measurements, and there’s some randomness and all this stuff. Those two attitudes, which both fit the data in some way, lead you in very different directions as to how to get better, right? How to build a bigger theory.

– [Brian] Yep.

– And sort of my comparative advantage as a working research physicist is that the rest of the world has not caught onto the Everett interpretation. So like, there’s all this low hanging fruit about using Everettian logic to address questions in cosmology and quantum gravity and things like that, that people just haven’t done.

– Sure.

– And so my grad students and I are having a merry little time, you know?

– And just say, “Hey guys, you can stay over there.”

– Right, yeah, like you know, remain confused for a while because I’m very slow, you know, and I need time to work this out. But I do think that, you know, to put it in more charitable terms, one big reason why our physics colleagues don’t care that much about the foundations of quantum mechanics is they don’t think it matters.

– [Brian] Yeah.

– They don’t think it changes or affects the research they do, the ideas they’re gonna come up with, et cetera. I think that they’re wrong.

– I agree.

– And the best way of convincing them that they’re wrong is for me to have a good idea, or for someone to have a good idea. I might inspire a generation of younger people to have good ideas, and show that thinking about things in this way leads to progress. That’s what I’m trying to do.

– Yeah, no, it’s funny, I was having a conversation with David Deutsch a couple of weeks ago and I noted to him in that, and I’ll just repeat myself here. A number of years ago, I was teaching undergraduate quantum mechanics for maybe the first time in a long time, and I was doing the whole year version as opposed to the semester version. So I was like, what do you put in the second semester? I thought that would be a good moment to introduce things like Many-Worlds, or de Broglie-Bohm, or GRW, just to give the students a sense of the various attempts that people have made to square the mathematics of quantum mechanics with our experience in everyday life and in the laboratory. I was really surprised at the pushback-

– Oh really? Yeah.

– That I got. Because the view, the general view is that’s a waste of the student’s time to be taken. They need to learn time dependent perturbation, you know, and I agree-

– Wow.

– That understanding the technical side of the subject, yeah, that is important, but you kind of need to grapple with what the theory is actually saying about the real world, so. But along the same lines, so you noted that all, I won’t be able to quote you exactly from a moment ago, but all the other ideas are really bad, you know, when you judge them. So, of course I know what you’re saying. The other ideas involve new mathematics or assumptions about things.

– Ad hoc ingredients.

– Ad hoc seeming things. The one example that has certainly caught my eye and not fully, but the de Broglie-Bohm approach, which it’s kind of this dark horse candidate that, you know, for various historical and sociological reasons, kind of was repressed over time.

– Yeah, that’s true.

– But this idea that, you know, Louis de Broglie and David Bohm rediscovered this notion that a particle like an electron can have a definite position, a definite speed. It’s still described in the probabilistic language of traditional quantum mechanics, but the probability wave does something a little bit different than in the conventional formulation. It plays a direct role in, pushing’s not quite the right word, but allow me to be a little bit loose, kind of pushing the particle around, pushing them toward locations where there’s a high probability, away from low probability. That is interesting to me because you have to give up the definite reality that we’re familiar with from Newton, but you don’t have to give up particles having definite locations and speeds. So it feels like you’re giving up less in this approach than in any other approach. I mean, did that have any pull on you too or?

– Nope. But I think, I mean, it would have-

– And tell me why.

– Yeah, it would have if the world had been described by non-relativistic point particle quantum mechanics.

– [Brian] Right.

– I think the single best sales pitch for de Broglie-Bohm is the following. Like if you were in the 1920s, and you were, you know, enjoying your Belgian ale at the Solvay Conference, and trying to understand the foundations of quantum mechanics and people said, “Well, sometimes the electron is acting like a wave and sometimes it’s acting like a particle.” And de Broglie comes along and says, “That’s because there’s both. There’s a wave and a particle.” That’s actually a very plausible move, except that then people invented quantum field theory.

– Yeah.

– And now all you have are fields, and the particles pop out.

– Yeah.

– For free.

– I agree, yeah.

– And so you don’t, like the motivating issue has gone away, I think, and at a more technical level, when you try to invent a version of de Broglie-Bohm theory that is compatible with modern quantum field theory, it’s not impossible, but oh my God, it looks bad.

– It’s pretty ugly. It is pretty ugly but I’ve always-

– But I would go even further to say that when you try to say space time is emergent, like that doesn’t play well with de Broglie-Bohm at all.

– I agree. I agree. And I’ve wondered, and I’ve heard adherence to this perspective claim, although I’ve never really seen it realized, that there are clean, straightforward ways that they could imagine creating a quantum field theory-like version of de Broglie-Bohm. So, and I’ve not delved into that enough to know whether that actually works, but it’s a brilliant idea, certainly in the non-relativistic scenario. Of course, the reverse argument is the one that you already made reference to. In the lean mathematical approach of Hugh Everett as developed by many others, including yourself in the decades since, we do have to embrace a reality that is so radically different from our experience in one single world. Now, I guess one can certainly say to that, “Hey, suck it up,” right? I mean, the world is not constructed for you to feel good, or to have a deep intuition at the outset.

– Right.

– You have to allow your intuition to follow where the theory or the mathematics or the data actually takes you. I mean, I presume that’s where you’re at on that.

– Yeah, 100%. I mean, and that’s why I really care about physicists caring about the foundations of quantum mechanics more than physicists caring about Everettian quantum mechanics.

– [Brian] Yeah, right.

– And I even have in my book, “Something Deeply Hidden,” I have this cute little abstract picture, which is like I draw some circles with some connections to them and I say, “This is the world of our experience.” And I say, look, there’s two ways to go. One is I have a theory that looks kind of like circles with some connections between them, and it’s very obvious how the theory matches reality.

– Right.

– Or I have this other theory, it’s just like one beautiful circle, and the connection to reality is very complicated, right?

– Complicated. Right.

– And so they’re both asking something of you and they’re both not immediately compelling. You need to work to make sense of them. And so the people whose intuitions are pushed in the direction of hidden variables, et cetera, I get it, I wish them luck. It’s not my thing, you know? Life is short, and whereas when I take Everett seriously and I really think about what a quantum theory is, and how it works, it’s like there’s a million questions that I get to answer, and we’re making progress on them, and it’s just so much fun that I’m gonna just wish them well.

– And is your intuition aligned with the cartoon version that we like to describe that, I mean, do you really think of the other Sean Carrolls out there in these other worlds and some having a better life, some having-

– I really do, and you know, I have the app on my phone, which will split the universe. I don’t know if you know about this.

– I don’t know about this one.

– Oh my God. The Universe Splitter app. It was-

– So it makes a measurement of some sort? Yeah, yeah, okay.

– Yeah, it sends a photon, photon to beam splitter.

– Photon, got it. Yeah.

– And it was only for iPhones. I think it’s now available for Android. I actually like helped inspire it.

– iPhone is good enough.

– So it’s-

– Are you getting royalties on this?

– I’m very proud of it.

– No. But I’ll push it anyway.

– Well, in one of the other worlds, you are, so.

– In one of the other worlds, maybe.

– Yeah.

– But, so I like to say that, you know, the wrong cartoon version of Everett is that when you make a decision, a new world appears, right? That’s not true. It’s when you entangle a quantum system with the environment, that’s when new worlds appears.

– [Brian] Right.

– But the other way around can work. So if you haven’t made your decision, you can ask the phone which one you want to do, and then yes, I really do believe-

– And then you really have to split the world.

– There will be, yeah, there’ll be one in which you do one thing and the other.

– And so this is a parlor trick that you pull out now and then.

– Yes, and it’s a crowd pleaser.

– Yes.

– Kids love it. But I’m increasingly of the opinion just as a sort of marketing thing that even though I have no problems with the phrase Many-Worlds as a description of the Everett approach, to quantum mechanics, the worlds are not the point. They come along-

– Of course. Yeah.

– And I think you should accept them because your theory predicts things. Until you get a better theory, you should accept what it predicts. But that’s not the essence of the theory. It’s not about the worlds, it’s about obeying the Schrodinger equation all the time.

– Yeah, right.

– And that’s where the questions come from.

– Right, but can’t you imagine… Well, of course the answer’s gonna be yes, we’ve already established.

– I can imagine lots of things.

– But you know, people like Gerard ‘t Hooft, you know, people who, Albert Einstein. Now, again, Albert Einstein was wrong about many things. He was right about many things. But it takes such a radical departure from reality, as you know, to fully accept the Many-Worlds approach that at least you can understand the motivation for developing versions that are closer to what we experience. Because, you know, it’s, again, easy to imagine, 100 years from now, people looking back and kind of chuckling and saying, “Those guys are actually thinking about all these worlds in this quantum wave function, when all you actually need is X,” you know.

– 100% I understand the motivation. I think that if I look at the history of physics, more often than not, the mistake has been on the side of not taking the implications of your theories seriously.

– True. Yeah, Steven Weinberg has some quote along those lines.

– Right.

– You know.

– And I actually never was familiar with this phrase before, but do you know about Kelvin’s paradox?

– No, I don’t know if I do.

– So you know back in the 1800s, they were developing thermodynamics. They discovered, oh my God, entropy increases. Doesn’t that eventually imply the heat death of the universe, right?

– Yep.

– And it’s a very short journey to go, because they didn’t know about general relativity, the big bang or whatever.

– Right.

– They lived in a Newtonian world that apparently would last forever.

– Right.

– Kelvin’s paradox is simply if it takes a finite time for the universe to equilibrate, and the universe is infinitely old, why hasn’t it equilibrated?

– [Brian] Yeah, I see, right.

– And so they knew that okay, you know, the sun is shining. It can’t go forever, et cetera. That’s what led Kelvin to do his famously wrong but, you know, good for his time estimate of the age of the sun and things like that.

– Yeah, yeah, yeah. Right, right.

– They could have said the universe had a beginning.

– Yeah.

– Right? And no one did, right?

– Right?

– And I think that’s the direction in which we tend to fail. We lose the courage of our theories.

– Yeah, no, no. I mean, certainly there are many instances of that in the history of science. But, so when you think about quantum mechanics and its need, for instance, to go beyond our current understanding, right? I mean, it is a theory that we’ve done extremely well to describe the electromagnetic force, the nuclear forces, the weak and the strong nuclear force. We’ve had a real headache in bringing gravity into that story. You made reference like string theories and approach to that. I don’t know if string theory is right. I never claim that it is. I claim it’s a good existence proof that quantum mechanics and gravity can play well together in a mathematical structure, and that to me is the real insight of string theory. It shows it can be done. It’s not necessarily the way it is to be done. Is there a deep lesson in there, along the lines of Kelvin’s mistake that we may not be paying sufficient attention to?

– I think that there might be. I don’t know, I can’t argue strongly there is, because we don’t understand quantum gravity, et cetera, so we don’t know what the reconciliation will be. But look, because I’m a never Everettian, and so my job is to think about the theory of everything. You know, what do you mean by theory of everything? Like ordinarily people would say, well, there’s some either particles, or fields, or strings, and they have some dynamics, and we’re gonna quantize that, and that’s gonna be the theory. Not for me, that’s not a theory. For me, the theory lives in the space of all possible quantum states. Hilbert space, as we call it, right? And talking about things like space, and particles, and all those are emergent, in the same sense as consciousness is emergent from the lower level.

– Sure.

– So you are forced to talk about physics in a very different language, right, and hopefully you can show that all this familiar stuff emerges from that. And so from that perspective when you say like, don’t give me some stuff and then quantize it. Just give me the quantum theory and then show me the stuff comes from it.

– [Brian] Right.

– That flips a switch-

– Sure. Yeah.

– In your mind, and you come to the point, I come to the point where I say like, look, like you just said, electromagnetism, the nuclear forces, matter fields all fit in well with quantum mechanics. Really what they fit in well with is the idea that you start with a classical theory and you quantize it.

– Yes, for sure.

– And I think that the difficulties we have with quantum gravity come about the fact that when we start with general relativity and we quantize it, it fails in various ways. And so the obvious thing to do, which everyone but me is trying to do is start with a better classical theory, which is basically what string theory is in some sense, right? You have some dynamics and whatever you quantize it.

– [Brian] Yep.

– I want to just start with a wave function that is not a wave function of anything, and ask if gravity can emerge from that. And look, the answer is maybe not, but people haven’t tried yet.

– Right.

– Which is just weird because we could have done this 50 years ago, right? And so I think that it might be that the reason why quantum gravity is hard is because we haven’t taken quantum mechanics seriously enough yet, and we’re still stuck with starting with classical theories and quantizing.

– Now, don’t you find, I mean, it feels to me that, again, not proselytizing for string theory here, but it feels like string theory goes partway toward that vision, right? Because as you said, the normal approach is take your classical theory of gravity, whatever it might be, try to overlay the quantum formalism in the manner that has worked when you overlay it on classical electromagnetism and the classical versions of the nuclear forces. String theory doesn’t do that. String theory starts with something where gravity has no apparent footing in the theory. It’s a vibrating string after all, and that’s all that it is.

– In space time.

– Yeah, so there is an environment. I totally agree with that. But let’s say you didn’t know that space time itself was the fabric of the gravitational theory. I mean, it took Einstein to get you there. So you have this vibrating string, you quantize this thing and out pops gravity.

– [Sean] Right.

– To me, that move, and I just feel like the critics of string theory and, you know, the folks on the internet-

– Some of our best friends.

– Some of our best friends who absorb some of the critiques that are made don’t fully appreciate the fact that out of the quantum vibrations of a filament emerges Einstein’s general theory of relativity.

– Right.

– Which that doesn’t mean it’s correct, but that’s kind of a mind-blowing set of developments.

– Yeah, and so I’ll say two things. One is this kind of sketch of a program that I’ve outlined, starting purely quantum mechanically with no stuff.

– Yeah.

– Just a quantum state in Hilbert space, and seeing how space and time and things like that emerge from it. It’s entirely possible that at the end of the day, what emerges is string theory.

– Sure.

– Or at least like part of string theory and, you know, whatever.

– A part of it, yeah, yeah.

– So it’s not incompatible. It’s coming from a different direction. And the second thing is I’m entirely on your side with being absolutely amazed at the success of what string theory, like the hoops through which string theory was able to jump. And I think that I’ll say that, you know, it’s not just that gravity emerges, but it emerges in a way without infinities, right?

– Yes. Exactly, yeah, yeah. Which was the whole headache.

– With all the matter in it. Like, you know, without, you know, any arbitrariness. And somehow despite the fact that, and you have done more than anybody, but we’ve tried very hard to let people know how exciting string theory is. We’ve still failed. I don’t know if you know, but we there was just a survey done by the APS, American Physical Society. Phil Halper and Afshordi organized it.

– [Brian] Yeah.

– They surveyed APS members on a bunch of controversial questions, foundations of quantum mechanics, what is making the universe accelerate?

– Yeah.

– Quantum gravity. And when it comes to quantum gravity, like no opinion, one, handily. String theory was the first non-trivial answer with like 19%. So what this means is among more or less professional physicists, around 19% of people thought that, not that it was right, but the leading candidate is string theory. 18% said that gravity might not be quantized.

– Really?

– Yeah, yeah, so you’re only 1% ahead of that.

– Wow. And that’s a very interesting statistic. I was not aware of it. You know, it’s funny, it brings to mind there was a article in “The New York Review of Books” many years ago by Freeman Dyson, who is a hero in the quantum world of physicists. And he was making the case that gravity and quantum mechanics need not come together. And so, you know, this patchwork quilt idea certainly simplifies things if you can describe the world, but it just doesn’t seem how that could possibly be the case. Right, if you have a quantum superposition, it gravitates.

– Yeah.

– And so what do you do with the fact that you’re in a quantum state that’s unlike what anybody like Newton would’ve ever put forward?

– Well, these days I do actually, I think I remember that Dyson article. There are more sophisticated attempts to do exactly that. People like Jonathan Oppenheimer have tried really hard to like write down theories with equations that have space time be classical, and something quantum on top. But my intuition is exactly yours that I think that the way it gets resolved is it’s not really quantum mechanics. It’s like modified versions of quantum mechanics to make everything work out. And that’s just, I mean, okay, but it’s, again, very ad hoc, very ugly. And I think that there’s just, you know, that life is gonna be simpler and more beautiful than that.

– Yeah, no, I can’t imagine that that was ultimately going to be the answer. So, when we began early on when talking about the laws of physics, you used the following language. Again, it’s not gonna be an exact quote. But you said, “I do believe that there’s stuff out there.”

– [Sean] Yeah.

– And so now we’ve gotten to a place where the fundamental architecture in the view that you are propounding is that, you know, it’s this abstract mathematical gadget that happens to have this name called Hilbert Space. It’s just the place where quantum wave functions naturally live. So when you think about stuff out there-

– Yeah.

– Are you thinking fundamentally there being Hilbert space, and stuff being this wave function that’s living in there?

– Yes, I really am. And I know, it rubs people the wrong way. Look, you know I’m a philosopher now too?

– Yes, I know.

– I used to be a physicist but now I’m in a philosophy department as well. And we like, we philosophers like to throw in some Latin occasionally, just to spice things up. So when people say, “But what is the universe in your view?” I like to say the universe is sui generis, Latin, for it’s its own thing.

– Yeah, right.

– It’s not a thing. It’s not like made of macaroni or whatever, right? There’s no other answer. And it’s not that the universe is living in Hilbert’s space or is a quantum wave function. Of course, those are the best mathematical representations of the universe. But the dramatic claim is that a quantum state living in Hilbert’s space could be an exact and complete representation of physical reality. And that might be a limit. Like, some version of that might actually just be true. And so whether or not it is, it’s plausible at the current state of knowledge-

– Sure.

– And my goodness, shouldn’t you pursue-

– Yeah, no, totally.

– That possibility? But it would also be completely insane that we human beings here in really the 20th century stumbled upon on this little planet, you know, in the outskirts of this galaxy, we stumbled upon the fundamental description of the entire reality. And moreover, it’s radically different from anything we see. It lives in, yeah.

– Right, yeah.

– So it could be true.

– And that I know it.

– And that you know it, yeah.

– Like that’s completely crazy.

– Yeah, yeah. And so does that give you pause? Or obviously you’re right. You gotta push forward with the best thing that we have here. But like, you know, do you have confidence, you know, presumably, maybe, maybe not. But imagine we make contact with alien civilizations in our lifetime, and they’re not so far away that we can’t have some conversation. If they’re advanced, do you think they’ll be like, “Yeah, it’s a Hilbert space.”

– Well, that’s the thing. I simultaneously believe like we have no right to think that, but also could be true, right?

– Yeah.

– So in other words, I do think that it’s a little bit different than previous episodes in the history of physics, because there were always things that were like manifestly not fitting the data, I would say-

– Yes. For sure.

– In earlier eras. And of course we have things like dark matter, the big bang that in some sense, we don’t have a theoretical explanation for, but they’re not incompatible with the idea that the universe is fundamentally a vector-

– We have many ideas that could accommodate them.

– In space, et cetera, right. So in other words, there’s no empirical pointer beyond-

– Yeah.

– That conception. So I think there’s an open possibility that, yeah, that’s just it, and we and the aliens are gonna be like Hilbert space, yeah. Who was your Hilbert?

– That’s right, that’s right. They’ll have a different name for it. So you know-

– Hilbert, of course, never named it Hilbert’s space. It was von Neumann who named it.

– Oh, is that true? I didn’t even know that history of that.

– Yeah, yeah, yeah.

– Yeah, yeah, no.

– John von Neumann, yeah.

– For sure. You know, so one of the criticisms of the Many-Worlds approach to quantum mechanics, which you know well because you’ve worked hard on this issue, is the way in which we became empirically convinced that quantum mechanics is a good description of the world is by making probabilistic predictions. No longer the electron will be here, but rather 32.2% chance the electron will be here, 15.7% chance here and so forth. And then we simply ran the experiments, identical version over and over and over. We collected data and we found that empirically, if it was a predicted 33.2%, whatever the number might be, we found it 33.2% of the times. So probabilities seem to be intrinsic for the reason we even believe this theory at all. Now in a Many-Worlds approach, you kind of need to rethink what you mean by probabilities because now every outcome is happening with probability one, because in some world, the electronic here and here and here and so forth. And so there’s been a long argument, discussion, deliberation, papers, you know, up the wazoo of trying to make sense of something that looks like probability in the context of a world in which fundamentally, there may not be any.

– Yeah.

– So where do you think this stands? I know you’re convinced that this is really now-

– That’s right, yeah.

– Solidly understood.

– Well, I think that I’m a subjectivist when it comes to probability. That is to say I think that the right way to think about probability is there’s something I don’t know, so I have various degrees of belief that I will attach to different possibilities. This is in contrast with people who believe that probability is an objective thing.

– [Brian] Yeah.

– And they will tell you the only real probabilities are I could roll the dice in principle an infinite number of times and get a fraction of them coming up two, which would be a sixth, right?

– [Brian] Right.

– But of course we use probability talk in all sorts of contexts. You know, the probability someone’s gonna win a football game, or win an election or whatever, where it’s not doable an infinite number of times.

– Sure.

– And that’s completely compatible with the subjectivist idea that probability is indicating a degree of belief. And so is the rolling of the dice. Like if you’re someone who believed in deterministic laws of physics, there’s a fact about where the dice will come up, but I don’t know the facts, so I assign it a probability.

– Right.

– So it’s all fine. I think that Everett is just like that, that Many-Worlds is just like that. When you measure something, you measure a spin and it’s either up or down, and so now there’s two worlds, one in which it’s up, one in which it’s down. There’s a empirical fact about the process by which that happens, which is it happens really fast. It happens so fast that there is a moment when the worlds are separate. There’s a version of you on the branch where the spin is up, another version of you on the branch where the spin is down, but you don’t yet know, okay? So in that moment, which inevitably happens, both of the copies of your former self that are on these two branches are uncertain about which branch they’re on. They have some subjective uncertainty about that. And then the question is, you know, there is, like you said, there’s a debate, a lot of papers written. Are you forced to assign probabilities just by the requirements of rationality? No, you are not. But is there like an obvious right way to assign probabilities? Yes, there is. And it turns out that even if you didn’t cook the books ahead of time, the obvious right way to assign probabilities is the one that works for quantum mechanics, what we call the Born rule, after Max Born.

– Yeah.

– Olivia Newton-John’s grandfather. And I think that it’s actually a little bit less arbitrary than it can seem. You know, I analogize it to, the technical phrase here is self-locating uncertainty.

– Telling you where you are in this, yeah.

– You might know the entire state of the universe except which one you are in.

– Right.

– So self-locating uncertainty. And like that’s, you know, it’s like big, weird metaphysical concept. But look, we don’t know whether the dark matter is let’s say a weakly interacting massive particle or an axion. These are two particle physics candidates where the dark matter could be, and there’s other candidates too, but let’s just simplify it to just those two. In some very real sense, there are two possible worlds. One possible world in which the dark matter is a weakly interacting massive particle, one in which it’s a WIMP. And if you say, “I think there’s a 40% chance it’s a WIMP,” right? Or axion, either one, you’re assigning some self-locating credence to these possible worlds. I think that all of science is that. If someone says, “Are you rationally required to believe that that 40% chance of it being an axion?” Like no, but there’s a better way to do it. There’s more room to play in scientific theory assignments, because people disagree about scientific theories, than there is in Many-Worlds.

– Right.

– In Many-Worlds, it’s so obvious what you should do.

– Right.

– And I think that as long as you just accommodate yourself, just like moral subjectivity, probability subjectivity too, like there’s a lot of stuff that is not out there in the world. We make it up, but it’s still really useful.

– And so in a way though, does that reduce the role of science and the role of the physicist? It’s basically taxonomy, right? It’s all about just figuring out, you know, the Library of Babel. So the example you gave earlier, is it basically all of our work is trying to figure out which world we’re in, which book on the shelf we actually pull off to describe ourselves?

– I mean-

– Are we just librarians? I mean-

– That is more or less exactly Carl Popper’s philosophy of science, right? You know, Popper famously said “We demarcate science from non-science by saying what is falsifiable.” And he did not like ordinary theory, confirmation theory that, you know, Bayesian reasoning and things like that. He basically said, “What you should do is invent every theory and then you should falsify the false ones one by one, and the thing that’ll be left is the right theory,” right? So yeah, in some sense, that’s what it is, again, for the worlds we’re on, but also for the laws of physics. If you think that laws of physics could have been otherwise-

– [Brian] Right.

– What we are doing as scientists is discovering what the actuality is.

– In this branch.

– In this space of, well-

– I mean-

– There’s two things going on.

– Yes, right.

– One is in the space of possible worlds-

– Yes.

– With different laws of physics. By doing physics, we discover our actual world.

– Right.

– In the space of branches, by looking around us-

– Sure.

– We discover which branch.

– Right.

– Same thing.

– And so does that, I mean, obviously doing science is about trying to gain insight into our world, would you say, insight into our branch. But in a reality where there’s so many variations on the world that we happen to see, does that in any way feel diminishing? I mean, to me, it kind of changes the game, the nature of the game. I mean, Einstein famously said, you know, “Could God have created the universe differently?” There’s something deeply compelling about a unique universe that we are just trying to understand, whereas if all possibilities compatible with whatever the foundational equation called Schrodinger’s equation, if they’re all out there, somehow it changes the nature of the game.

– I can do no better than to quote our old friend Joe Polchinski.

– Yeah.

– Where he said, “I never promised you a rose garden.” Yeah, I mean, you got to remember, Many-Worlds doesn’t say every possible world happens, right?

– Sure, it’s got to be compatible with the-

– Gotta be compatible, blah, blah, blah. And, but many, many, many worlds happen, and yeah, that’s what it’s like. And again, our job is not to impose our desires on the universe, but to try to figure out the best account we can make of the universe we actually find ourselves in.

– Yeah, yeah, for sure. Final topic, if you still have enough energy to continue, you know, you made reference earlier to heat death of the universe, second law of thermodynamics, you know, Kelvin’s mistake and so forth. In recent years, and I know that you’ve spent some time thinking about this, we’ve paid more attention to not just entropy, you know, this measure of disorder. It’s a word that, I don’t know, for some reason people find it off-putting, but it’s really just sort of this measure of the amount of organization or lack thereof. There’s this other notion called complexity-

– [Sean] Right.

– That is very useful to introduce to try to understand the actual structures that we see in the world. Can you just give-

– Yeah.

– A feel for what insight this idea brings that entropy in the second law just glosses over in some sense?

– So I do think that complexity is a wonderful scientific concept, in part because I would argue… Some of my complexity friends disagree. I would argue this pre-paradigmatic. And you know, Thomas Kuhn had this idea that when a new science comes around, you don’t know the rules of the game, right? Like you don’t know what are the important examples, what are the important equations, what are the questions you’re asking, and you’re just sort of fumbling around like, you know, Galileo and Descartes were trying their best with classical mechanics, and then Newton comes along and gives you the paradigm, and now you have questions that we all agree are important and we’re all pushing forward in the right way. I think complexity is pre-paradigmatic. We don’t have an agreed upon set of the most important things. There exist textbooks on complexity science, and they all have a grab bag of, you know, different things. If this is chapter eight, we’re talking about economics, right? So the hope is, the aspiration, and I think it’s, again, completely gonna be a matter for reality to decide. We can’t decide ahead of time ourselves. Are there features of complexity as such that are common to economics, and computers, and biology, and consciousness enough that we can usefully use these tools? And I think that there’s a lot of evidence that the answer is yes. I mean, roughly speaking, there’s different notions of complexity, different definitions, just like entropy.

– [Brian] Sure.

– The very simplest notion of complexity is how much information would you need to give me to describe a system?

– Yep.

– Okay. So, and that sounds like how big the system is, but that’s not true. The integers are infinitely big. There’s infinitely many of them.

– Very easy to describe them.

– But they’re very easy to describe, right? Whereas the United States is smaller than the integers, but you need a lot of information to capture it, right? So that’s one notion of complexity. But there’s also notions that are more functional, and as a quasi philosopher, as a sort of fake philosopher, they rub me the wrong way because what’s a function? You know, what’s a goal? What’s a teleology? And people invoke these words fearlessly, and I’m a little fearful of them.

– [Brian] Yeah.

– If one, so one definition of complexity I heard is, you know, a car, by this definition, a car is not complex. Complexity happens, complexity doesn’t happen, complicatedness happens.

– Yeah.

– And you have many different pieces that come together, but each piece has one role. Complexity happens when you have many different pieces that don’t start off having different roles, but they come together and take on different tasks for the greater thing, right? Like the cells in your body start off similar. The people in a society start off similar, right? The starlings in a flock or whatever. But you get this complex emergent behavior out of the interactions between the pieces. So I think there’s something true and real in all these different conceptions. But the question just like with Everettian quantum mechanics is show me the money. What are you gonna do with this, right? Like what are you gonna actually learn from it? My favorite example of something we actually learned comes from Geoffrey West and his collaborators at the Santa Fe Institute. And he said, look, there are these well-known things called, the technical term, I know the audience likes to get some buzzwords for cocktail parties, allometric scaling relations. This is a relationship in, let’s say mammals, for example, as a group of animals between their mass, their metabolic rate, so their heartbeat, and their lifespan, okay? So you can like have all these different variables, and they turn out to be tightly correlated. And something that has been noticed since the 1930s or something like that is not only are they correlated, but they’re correlated using what are called power laws. Like your metabolic rate is roughly, this is all very rough, right, but in a species of mammal is your mass to some power, and your lifespan is your mass to some power. And you can plot then, you know, show the plots and it’s great. It’s beautiful. Like for biology, like it fits the data really well, right? And then someone else noticed, okay, they’re not only mass to some power, the powers are all like one fourth, or minus a fourth, or three fourths or something like that. And sometimes they cancel out. So if you ask the number of heartbeats or the heart rate of a mammal and it’s mass, that does something, and then its lifespan also does something, and you multiply them, every mammal gets one and a half billion heartbeats in its life, right? And this is the kind of thing that calls out for an explanation, like that can’t just be random. And so West and his collaborators explained it. They came up with a theory based on bifurcating networks that end in nodes that have constant energy rates, and that one fourth is one over the dimensionality of space plus one.

– Really?

– Yeah.

– That’s very nice.

– If we were in four-dimensional space, it’d be one fifth.

– It’d be one fifth.

– Exactly. Right. And it’s a, is it predictive? Well, maybe not. I mean, you can predict things you haven’t measured yet, but it’s explanatory in a really nice way. And so I think that that will be, hopefully the aspiration is that’s the beginning of a paradigm, right? Like here’s a set of ideas, networks, power laws, things like that, that can grow into an explanation of, because power laws are everywhere. The internet, your brain, things like that. And so I think that there is something to the idea that when you have many little pieces working together in concert to have some higher-level emergent complex behavior, there’s certain ways that generally robustly work and certain that don’t, and that will be the study of complexity.

– Yeah, so that’s a rich, possibly rich subject.

– Absolutely.

– And I said it with the last question, actually, one more that just I think it’s worth spending a moment on before we wrap up. We spoke a little earlier about artificial intelligence, LLMs, but really in the context of consciousness.

– Yeah.

– When you think about the future of physics, and the role that these systems may play in advancing our understanding perhaps more quickly, more effectively, more efficiently, I’ve had some conversations with some of our colleagues who have said things like, “Pick your final problems.” Because you know, five years from now, these systems-

– Right.

– And my own experience is somewhat limited. You know, I wrote a paper recently. You made reference earlier before we started to Janna Levin. Janna and I know, you know, Massimo Porrati and Dan Kabat, we wrote a paper. And I was wondering how long would it take me if I treated Chat like a grad student to get it to the result? And it wasn’t long. It was like a really good graduate student that could sort of get right there. And it’s pretty early game, right?

– Right.

– Early on in the… So what do you think about the future of physics research in an era of powerful AI?

– I do think it’ll be transformative. I do think that it will not wipe us out. Like you don’t need to pick your last problem. You have at least three or four problems.

– There it is.

– I think that’s right.

– Well, that’s even just in my lifetime.

– In your lifetime.

– So there you go on that right there.

– I don’t want to speak for your students. I don’t want to be them. But I had a experience at Johns Hopkins where a postdoc gave a lunchtime talk-

– Yeah.

– Where he did the following thing. And he was someone who was in the weeds of the data in gravitational waves. So he had some data set from LIGO, the Gravitational-Wave Observatory, two black holes in spiraling. What you would typically do is you would study the system, you would analyze, you would get the period and whatever, you would write a paper, right?

– [Brian] Yeah, sure.

– He instead wrote a prompt. It was about a page long, and he had the data on his computer, and he had a LLM on his computer. At the beginning of his talk, he submitted the prompt. An hour later, the paper was written.

– Oh, Jesus Christ.

– Not just the analysis was done. So the paper with references, with plots, with the whole thing.

– And this was fresh data that the system-

– Yeah, had not been analyzed before.

– Had not been ingested before.

– Now, of course you’re gonna want to check this because they make mistakes, right, okay.

– [Brian] Yeah.

– But I think I’m not that, I’m impressed by it, but I’m not that surprised by it in the sense that that’s a closed-

– It’s algorithmic.

– It’s algorithmic, right.

– Right. Right.

– And so the dream, the aspiration is LLMs will make all the hard stuff go away.

– Right.

– All the boring stuff I should say.

– Yeah, the drudgery work.

– The drudgery work, right.

– Yeah, and that would leave us talented, creative human beings to do the real work.

– I mean, tone of voice and all that, but yes, in real sense. I mentioned I was just, I use LLMs in my work. They are, as friends of mine put it, an accelerator. I would never cut and paste from an LLM output into a paper.

– Well, you just can’t trust it yet, right.

– I don’t know if you saw, but literally the archive, our place where we put all of our science reports-

– No, there’s been a debate about how to treat submissions that haven’t, yeah.

– Well, they came up with a policy that if you have even one hallucinated reference in your paper, you’re banned for a year from submitting to the archive, so.

– That’s a pretty low bar.

– You would think. But we’ve had people fired at places like “New York Times” and “Ars Technica” because they’re grown up reporters who-

– Just making stuff up, yeah.

– Well, they use-

– They shortcut and-

– They shortcut, right.

– Yeah. All right.

– However, it’s really good for learning things like that are already understood by somebody else-

– Sure.

– And you don’t. You can ask it questions, and you can even dig in, right?

– [Brian] Yeah.

– But I was trying, like I’m doing a research paper, and by the nature of the kind of research I do, it hasn’t been done before, right?

– We hope.

– Right, you know, it’s close. It builds on things.

– Yeah, yeah.

– But because it hasn’t been done before, like I was asking, you know, the LLM to do a certain, explain why a certain function had a certain property.

– Right.

– Like a very standard thing. And it was wrong over and over and over again, right?

– Sure.

– And it was self-contradictory in its own things because it was just a little bit-

– And it apologized a few times.

– And it apologized that, says, “Yes, I was contradicting myself.”

– Yeah, right.

– And then it would contradict itself again.

– Yeah, and what do you use, which-

– I use either Claude or ChatGPT. I go back and forth. They’re good for different things. But like you say, the state of the art is improving and that will improve. But I still think at the end of the day, I can easily see a world in which LLMs could have solved Einstein’s equation to find the short shield solution for the gravitational field of the sun.

– [Brian] Right.

– I don’t think they would’ve invented general relativity.

– I do wonder about that. You know, look, post-facto, we can always tell interesting stories that make it feel inevitable.

– Yeah.

– Right?

– And so, you know, when I think about general relativity, I use the rigidly rotating disc to try to see how curvature would naturally appear. And once you have curvature in the story, if you’re really smart as the LLMs are, you bring in the architecture, the mathematical architecture. Of course, you got the Riemann tensor and the Ricci tensor. You got the metric in there. And now you’re knocking on the door of general relativity. So I really do wonder, this is a very hard experiment to do because the systems are already infected by the history of the subject, and they’ve ingested everything. But I would love to do an experiment where you train the system and you carefully excise any reference, say, to general relativity. Could it get it?

– So they’ve done that.

– Oh, they have done that.

– Yeah.

– Oh, I didn’t know that.

– And it’s kind of hilarious. The LLMs trained on, let’s say, data pre, let’s say you give it data that only existed-

– Before 1905 or something.

– Before 1913, or something like that.

– Yeah, right.

– Let’s say 1913.

– Okay.

– LLMs don’t believe in special relativity because you and I forget that people didn’t believe in special relativity.

– Oh my God, I know that well.

– And so-

– The anti-Semites especially didn’t believe. Yes, yeah, yeah.

– Well, yes, there was that. But it was all like, now we know that it’s established, we tell ourselves this rational reconstruction of the story.

– [Brian] Yeah, yeah.

– But you know, yeah, they know that it exists because it’s in their training data.

– And they were able to, they were able to excise in a convincing way.

– It’s not that you give it the whole thing and then remove post 1913. You just only give it things.

– That were never published by that.

– That’s right.

– And then how well did it, I mean, they must have pushed it.

– Well, as far as I can tell, they don’t think it does very well at pushing.

– Really? Interesting.

– But I do think, but look it might be, like again, I don’t know. Like I would’ve been wrong five years ago if you asked me about how good they would be now. But it’s not just that they’re good at reasoning or bad at reasoning. It’s that they’re good at a certain type of reasoning. They’re super good at interpolating between existing facts. They’re super good at remixing and pastiching facts. Like explain the standard model of particle physics in a series of 100 haiku.

– Right.

– They could do that better than me.

– Yeah, really can. It’s amazing.

– But it’s much harder to go beyond.

– Right, right.

– Extrapolating as hard.

– Yeah, so we’ll see. We’ll see where it goes. But it’s certainly exciting, and it does raise the question, what will happen to the need for graduate students as we go forward? I mean, we’re meant to be training the next generation, but if we’re more selfish, and we’re just trying to push our own agenda, research agenda forward, it may be that that whole structure that sociologically has been really important may be strained.

– I do think that, and I don’t think that I’m just being, telling myself stories to make myself feel better, but maybe I am. I think the human beings are always gonna be crucial. Always in the next 100 years, let’s say. I do think that even the rotating disc, you have to think of the question to ask.

– I agree. I agree.

– You have to like do that. And I do think that it’s the LLMs, we anthroporphize them and they sound human because they’re trained on human speech.

– Of course, yeah.

– But the process that gets them there is fundamentally different. So what I do believe is that maybe someday, we’ll have different versions of AIs that are not LLMs, right? That are not just learning models.

– Yeah, sure.

– That up close, more closely approximate the human thought process, and those might put us out of business but that’s not-

– Or in a completely different methodology that just blows us all away.

– Right, even better than what we do, right. That’s right.

– All right, well, in any event, it’s an interesting future. So thanks so much for the conversation and joining us today.

– Thanks, Brian. Yeah, thanks.

– Thank you.