The arrow of time — all the ways in which the past differs from the future — is a fascinating subject because it connects everyday phenomena (memory, aging, cause and effect) to deep questions in physics and philosophy. At its heart is the fact that entropy increases over time, which in turn can be traced to special conditions in the early universe. David Wallace is one of the world’s leading philosophers working on the foundations of physics, including space and time as well as quantum mechanics. We talk about how increasing entropy gives rise to the arrow of time, and what it is about the early universe that makes this happen. Then we cannot help but connecting this story to features of the Many-Worlds (Everett) interpretation of quantum mechanics.
Support Mindscape on Patreon.
David Wallace received a D.Phil. in Physics and a D.Phil. in Philosophy from Oxford University. He is currently W.A. Mellon Professor of Philosophy of Science, with joint appointments in the Philosophy Department and the Department of History and Philosophy of Science, at the University of Pittsburgh. He is the author of The Emergent Multiverse: Quantum Theory According to the Everett Interpretation. Among his honors are the Lakatos Award for outstanding contribution to the philosophy of science. His most recent book is Philosophy of Physics: A Very Short Introduction.
[accordion clicktoclose=”true”][accordion-item tag=”p” state=closed title=”Click to Show Episode Transcript”]Click above to close.
0:00:00.0 Sean Carroll: Hello, everyone. Welcome to the Mindscape Podcast. I am your host, Sean Carroll. The passage of time, it gets to us all, doesn’t it? I’ve been doing this podcast for over three years now. I remember almost every one that I’ve done. Actually, I remember every one, but I don’t remember any of the ones that I haven’t yet done. There’s an asymmetry, right, in my memory between the podcasts that I’ve already done in the past and the podcasts yet to come in the future, not because there aren’t any in the future, but because of some feature of the way that time works in our universe, the arrow of time pointing from the past to the future, that gives it that imbalance. And this is one of my favorite topics to think about in physics and philosophy, and the amazing thing is, we haven’t really talked about it here on the podcast yet.
0:00:45.1 SC: We did talk about time, the psychology and neuroscience of time with Dean Buonomano, but the physics and philosophy of the arrow of time, which was the subject of my first trade book, From Eternity to Here, we haven’t talked about yet. Also surprising is that we have not yet had David Wallace on the podcast. David is one of the leading philosophers of physics in the world today. The third surprising thing is that we’re having David on and we’re not talking that much about the many worlds interpretation of quantum mechanics. David is most famous for being probably the person who has thought most carefully about how many worlds should work in the real world, and I’ll encourage you to check out his book, The Emergent Multiverse, if you really want to dig into the details.
0:01:28.6 SC: But David is broad in his interests in physics and philosophy, and he’s certainly thought a lot about entropy and the arrow of time and the initial conditions of the universe and so forth. That’s what we’re going to be mostly talking about today. How are all these different things related in the typical, good philosopher, very, very careful way of teasing out what all the assumptions are, how they go into things. I know that I’ve found by talking to people that the arrow of time is one of the hardest things to think about in a reasonable way, just because this idea that the past flows into the future is so deeply embedded in how we think about the world that the idea that it’s something that has to be explained and that you’re not allowed to cheat [chuckle] when you explain it, is a hard one to wrap your mind around, but David is very, very good at it.
0:02:18.0 SC: And guess what? By the end of the podcast, we are in fact talking about the Everett many worlds interpretation of quantum mechanics, because it has a relationship to the arrow of time. The universe branches into multiple copies of itself going forward in time, not going backwards. So I think this is a classic upcoming episode here of the Mindscape Podcast and you’re all going to enjoy it. So let’s go.
[music]
0:02:57.1 SC: David Wallace, welcome to the Mindscape Podcast.
0:03:00.2 David Wallace: Great to be here.
0:03:00.9 SC: You know, we’re… We’ve had some philosophers on the podcast before. In fact, philosophers of science, in fact, philosophers who started out as physicists and then turned into science, like our friend David Albert, for example. But nevertheless, I thought it’d be fun to start by just asking you about that transition very quickly before we get into the meat and potatoes of the thing. What is it that causes a person to sort of at one time in their lives, think, “I want to be a physicist,” and then some other time, think, “I’d really be happier doing philosophy.”
0:03:33.8 DW: Well, I think there’s a bunch of people who get into physics and then gradually start thinking that a more philosophical way of exploring things is more interesting and they want to move in that direction. I’m not that kind of person at all. I still want to do physics. Largely, I still think I do do physics. [chuckle] The kind of work I do is interdisciplinary and I think relatively mundane facts about the way Britain and America organize their universities mean that for the kind of physics or interdisciplinary stuff I was interested in, staying in a physics department and trying to do that work wouldn’t have been the smartest career move.
0:04:11.6 SC: Right. Yeah.
0:04:12.1 DW: So I kind of finished my PhD in physics with a bunch of research interests that were kind of in various foundational things, quantum field theory, the Everett interpretation. And it looked pretty clear that I’d probably be able to get post-docs that way relatively easily, because everyone says they love it, interdisciplinary work, but when I started looking for tenure track positions, I’d be short of luck.
0:04:33.8 SC: Right.
0:04:34.8 DW: So it was… It was a fairly practical career move. I don’t want to understate how cool academic philosophy is or say that I haven’t developed lots of interests beyond that through exposure to it, ’cause I really have, but it really did come down to that kind of, I want to do this kind of physics. Where can I do it? How can I do it?
0:04:50.8 SC: I mean, it is a indictment, I think, of the academic system, at least in the UK and the USA, which is the ones we’re familiar with, that the siloing is strong enough, that there’s… There isn’t a spectrum in between these, right? You are either a philosopher or a physicist because you need to get paid. You need to get a job. And that affects how people choose their research topics and things like that.
0:05:16.2 DW: Yeah. Yeah. And people like interdisciplinary work in principle, but when it starts to be, you want a permanent position, you have a teaching component, it gets hard. And I do think it’s a genuinely hard problem to solve. I mean, one of the reasons foundational philosophical work’s tricky to do in a physics department is the criteria to measure it are hard. So you’ve done an experiment, the experiment does what you said it did. You’ve done a calculation, the calculation can be checked. All of that’s fine. Harder to do that with foundational stuff? I mean, and yes, as we both know, some genuinely bad stuff gets done in foundational spaces, which doesn’t get caught, by those kind of filters.
0:05:52.4 DW: So I think it is an indictment, but I also do think it’s a tricky problem to solve. And maybe one reason the philosophy of physics works quite well as a space is philosophy is a bit better in knowing how to assess for quality, that kind of more conceptual-oriented work.
0:06:05.4 SC: That makes perfect sense. And actually it leads very well into the meat and potatoes here, because my own… And I think I’ve said this before, but maybe not to you, I was always interested in philosophy ever since I was an undergraduate, but the kinds of philosophy that I liked were actually moral and political philosophy, because the philosophy of science that I learned was all about theory choice and sociology of science and stuff like that, which is important and interesting, but that’s not my bag. It wasn’t until I was a grown cosmologist and thinking about the early universe that I came across a bunch of philosophers who were really, like you say, really doing physics, right? People like Huw Price and David Albert and Tim Baldwin and yourself who were trying to understand how the universe works.
0:06:47.4 SC: And it was through… So it was through the arrow of time that I actually got interested in philosophical foundational questions. And so I thought it would be fun to start talking about that before we get into the Everett interpretation. So…
0:07:00.4 DW: Sounds great.
0:07:04.3 SC: The phrase “the arrow of time,” something people have heard, but then there’s this feeling that there are many arrows of time. What is your take on, is there an arrow of time, are there many, what are they? How do you define them?
0:07:13.9 DW: The basic way I’d started getting at it is something about which processes in the world run forwards and backwards just as happily, and which processes don’t? So if you look at… If you look at small systems, or not really even just at small systems, but if you look at systems with not many moving parts, in physics, they don’t seem to care whether you run them forward or backwards. So if I showed you a video of the solar system, I speeded it up a bit, and I asked you whether I was running it forward or backward, I mean, if you paid attention to whether the sun rose in the east or the west or something, you could probably figure it out, but it wouldn’t be obvious. The solar system’s physics doesn’t really mind if you run it forward or backward, and that’s just the same reason why, the way we work out when the next eclipse is going to be, it’s just the same way we work out when the eclipses were in the classical era, so we can cross-reference it to Thucydides or something.
0:08:03.0 DW: Most processes aren’t like that, processes with lots of moving parts aren’t like that. If I show you a speeded up video of ice cube in water or the way I’ve got older over the last 10 years, or a building collapsing or something, it’s completely obvious whether I am running a video forward or backward. So most processes in the world really distinguish between forward and backward dynamics, and yet big things are made out of small things, so there’s something extremely puzzling, kind of in the vicinity of a contradiction, if you’re not careful, with the idea that big processes clearly distinguish forward and backward in their dynamics, but they’re made out of processes that themselves don’t distinguish it.
0:08:48.2 SC: Do you yourself like to rigorously categorize different arrows of time or do you just lump them together as an arrow of time and different aspects of it?
0:08:58.4 DW: Yeah, I’m kind of skeptical about a lot of that categorizing of different arrows. I think… You know, people do talk about a statistical thermodynamic arrow and a cosmological arrow and a memory arrow, and I think… The way you break those things up kind of presupposes a structure that’s not necessarily there. I don’t think we should necessarily take it as read the direction of time implied by the fact that we remember the past and not the future, it has the same origin as would be assumed as the direction we get from… For the fact that ice melts and doesn’t re-freeze, but if we just kind of silo them off, then we can kind of assume that we know how those things break up in a way that isn’t completely reliable.
0:09:44.1 DW: And I think when we’re seeing that actually is you start… If you want to check if two things are logically independent, you want to try switching one of them off and leaving the other one on, and seeing if it makes sense, and if you try to imagine universes where you turn around, let’s say, the direction in which you remember things or the direction in which radiation decays, but you don’t turn around the direction in which ice melts, you very quickly have something you can’t really imagine when you start trying to think through the details in any length.
0:10:11.6 SC: So in other words, just to sort of leap ahead to a big claim that we’ll interrogate later, you seem to be saying that in your view, there is one underlying thing that ultimately is going to account for both why ice melts but doesn’t un-melt and also for why we remember the past, but not the future.
0:10:28.9 DW: Well, look, I reckon that’s true, but I’ll do a slightly more, sort of drawn back version for that. I’m pretty sure that the reason… If there is a relation between those two, then the direction of time in human cognition and action is what it is because of the direction of time in physics and not vice versa. So we ought to be able to figure out what’s going on in the direction of time in physics without having to presuppose a bunch of stuff from human cognition and action. If… Once we’ve done that, then I kind of strongly suspect that because humans are physical systems, it’s going to turn out that asymmetries in humans are a special case of these other asymmetries. But at the very least, we can figure out those asymmetries in physics without having to get tangled up in these issues about asymmetries in us.
0:11:17.0 SC: But this… So this does presuppose, and you’re in a friendly crowd here, so it’s okay, a kind of physicalism, right? We’re not attributing human agency to something over and above the fact that we’re made of stuff, so if we understand the asymmetries of the behavior of the stuff, the asymmetries of human thought and cognition should follow along.
0:11:38.1 DW: Right, yeah, absolutely. There’s a substantial assumption going on here, and there’s a sense in which I’m not even seriously defending it, it’s a bit too foundational to start defending. You might kind of draw it up like this, there’s this kind of approach I think you’re calling physicalism, some places I’ve just called it like dynamics first, that says something like, okay, the way we manage our science is… It’s a dynamical description of what’s going on independent of us, insofar as we interact with it that matters, ’cause that’s how we test stuff and because we want to understand us too, but the physics itself can be understood without consideration of the agent’s role.
0:12:14.1 DW: And there’s a completely different tradition that says, well, theories can’t be understood except insofar as they are things to be used by theorizers, and the role of the theorizer can never be taken out completely. So you’ll never really understand the asymmetry of time in physics without understanding there’s an asymmetry in how we as agents get at systems. Look, some smart, reasonable people take that approach, and sometimes it bears fruit, and there comes a time where I think it’s less useful to just go round and round in ever-decreasing circles, try to persuade somebody who has that starting point that they’re wrong, and it makes more sense to pursue your own project and have it judged by its fruits.
0:12:57.5 SC: That’s right. If you succeed, then the good things will flow in your direction, right.
0:13:01.8 DW: Yeah, exactly.
0:13:03.3 SC: Okay, so you mentioned, I think very correctly, that naively, this asymmetry of time and evolution of big complicated things might be thought of as intention or even contradictory to the symmetry of time and evolution in things with a small number of moving pieces. Before we explain why, do you know much about the history? Who’s the first person who pointed out this was a puzzle? [chuckle]
0:13:27.0 DW: So I should know this. Lots of philosophers of physics are really good at the history of physics. I’m not one of those philosophers of physics, but I hang out with them a lot, so I can do my attempted second-hand version. Back in the late 19th century, when Boltzmann was first trying to write down mechanical theories of gases, these ideas were in the air and developing, and there was clearly a lot of confusion. And it’s one of these quite delicate things to tease out exactly who knew what when, but roughly speaking, Boltzmann had a stab at doing an entirely mechanical explanation of why gases expand to fill the space they’re in, and why they approach an equilibrium, a kind of stable, well-behaved distribution of a certain kind.
0:14:07.0 DW: And he kind of, at least on the natural reading of it, he wrote a paper that tried to disprove this mechanically, and then a bunch of people, Zermelo, Loschmidt, pointed out like, “This is logically impossible.”
[laughter]
0:14:19.0 DW: You started with some time reversible equations, you made some time reversible assumptions about those equations, and you got out something which distinguishes the past and the future. This can’t be done. And also you started with… This is a slightly subtler point, but you started with some equations which reliably say that wherever a system starts off it will eventually get back there, and you’ve derived the fact that the system will start somewhere, go somewhere else and stay there forever. This cannot happen.
[chuckle]
0:14:41.8 SC: Right.
0:14:42.4 DW: So I think really very early on, before statistical mechanics was a properly developed discipline, right at the very first attempts to do this stuff, then people were very quickly seeing that there was an issue there. I don’t think it was much before Boltzmann, because you had to have a realistic and detailed understanding of how the large came out of the small to get the worry off the ground. I think Newton kind of liked the idea that there might be like an atomistic basis for everything, but it was so embryonic at that point that I don’t think anyone that I’m aware of seriously tried to think through what possible issues were there.
0:15:19.0 SC: Well, I guess one of the reasons I’m asking you is just because I don’t know the answer, and I’ve kind of wondered about this. I mean, I don’t want to say too many things rather than asking you to say them, but there is the Second Law of Thermodynamics, which roughly says that entropy increases, and we can get into that, but it predates Boltzmann, like Boltzmann gave an explanation for it in terms of atoms and molecules bumping into each other. So I’ve always wondered, once they did come up, Carnot and Clausius and people like that, they invented the second law, manifestly not time reversible, right? I mean, entropy increases with time, it doesn’t decrease.
0:15:55.5 DW: Yeah.
0:15:55.8 SC: And they also knew that there were Newton’s laws of motion, like that was well-known. And so did anyone… Was the idea that things like the second law were over and above Newton’s laws of motion, or was there even a hope of deriving them from the start? And maybe the answer is you don’t know, ’cause I certainly don’t know, and I’ve definitely tried to find out.
0:16:17.1 DW: The sense I have is that the apparent problem there just doesn’t bite unless you’ve got a kind of broadly, in the good sense, reductionist picture of what’s going on in the world. If you think that Newton’s laws are supposed to be laws for everything, then you can see the contradiction’s going to turn out pretty fast. If you think Newton’s laws is a great for planets, and cannonballs, and they’re not great for living things, or for clouds, then you’re not going to be so bothered, there’ll be bits… You have a fundamentally pluralist world you’re only beginning to understand, and you don’t know how all the bits connect together, and the fact that the bit that describes small mechanical stuff has a symmetry in time, and the bit that describes steam engines doesn’t have a symmetry in time isn’t going to bother you.
0:17:03.2 DW: And we can look back now and think how inevitable it ought to be that these things get put together, but you’ve got to remember, even in the early 20th century people were still taking extremely seriously the idea that living tissue was not described by the same laws of physics that describe non-living matter, that there were ground-level laws of nature that applied to living tissues that didn’t apply to non-living tissue. And it wasn’t even a stupid thing to think.
0:17:26.1 SC: No, no, not at all.
0:17:26.9 DW: It’s always a little unimaginative, but they didn’t know about quantum mechanics. So I think the discovery that the world actually is really pretty unified, and that in the good sense it’s pretty reductionist is a really substantive, important discovery of 20th century science. We don’t want to just read it back as some automatic a priori fact about science that they could have known just by cognition in the 19th century. So that’s my sense of it. It’s only when Boltzmann actually writes down something that starts looking like a proper mechanical description of steam, or gases, or at a bit more earlier than that when Maxwell’s starting to do similar things, it’s only once people are starting to actually put in the connective tissue between mechanics and thermodynamics, and starting to get results that make it plausible that they’re connected in this way that you start seeing the bite of the tension between the… That’s my sense of it, but again, it’s historically not super informed.
0:18:21.5 SC: No, but actually, it makes a lot of sense, and it… I like it because, if anything, it elevates the accomplishment of Maxwell and Boltzmann and their friends. It’s not just that they helped establish statistical mechanics, but reductionism, or even the unity of science in some sense. The idea that human beings are made of the same stuff and obey the same laws that atoms do is exactly, as you say, whoever came up with it, it’s a non-trivial step.
0:18:47.9 DW: Yeah.
0:18:49.6 SC: And so then this leads us into what Boltzmann did. What was the insight into entropy and disorderliness, and time that Boltzmann and his friends came up with?
0:19:03.5 DW: Well, again, I’m abstracting a little bit from how that kind of very messy history turned out…
0:19:07.9 SC: Right, the history is not a primary goal here.
0:19:10.4 DW: What they kind of started to realize was something like, okay, it’s not… It can’t be that you could just logically derive the asymmetries, the expansion of gases, the second law, in your example, from these mechanics, you have to put some additional assumption in. And the additional assumptions they put in, or at least one of them, one of the ones that gets a lot of attraction, was some notion of probability. It was some idea that, okay, the gas can’t… You can’t be sure that the gas will expand to fill the box, but it’s extremely likely it’s going to.
0:19:47.9 DW: And to some extent, that notion of extremely likely gets cashed out combinatorically, like how many ways could the gas be? And almost all the ways the gas could be in this cashing out are ways which it’s going to expand. That’s kind of a bit dodgy because there’s infinitely many ways the gas could be, and so comparing the numbers is questionable. But that kind of basic idea was there, that what was going to solve the problem was going to be probability. And the funny thing is, that lesson, actually, that idea that you get round this paradox by introducing probability is still, I think, quite often what you run into in the kind of second year undergraduate statistical mechanics course that wants to spend the first lecture doing the history of the subject, so it could spend lecture 2-16 teaching you how to calculate partition functions.
0:20:36.4 DW: But you step back for just a moment, and of course it can’t be true by itself, because leaving aside entirely some really deep questions about how we understand probability in this context, the mere introduction of probability doesn’t break the time reversibility. If you want to say it’s overwhelmingly probable that the gas is going to evolve to fill the box in the future, but you don’t want to say it’s overwhelmingly probable that the gas is going to have been such that in the past, it filled the box, then in some way you must have broken the time asymmetry, and so the time symmetry. And appealing to probability by itself doesn’t change that paradox at all.
0:21:18.6 SC: And actually, I want to dwell on that statement because it was beautifully put, and it’s at the heart of a lot of difficulty that people have with the arrow of time and with related notions. Because they come up with some statements, some purported explanations that all sound good, but even though what they’re trying to explain is the asymmetry of time, they don’t go through the exercise of running that thing backward in time and seeing if it sounds just as good. [chuckle]
0:21:45.1 DW: Right, absolutely. I have an article I did this a while ago, that said, “Computer scientists say garbage in, garbage out. If your input is a mess, your output’s a mess, no matter however clever your program is.” The slogan in statistical mechanics or philosophy of statistical mechanics ought to be, “Reversibility in, reversibility out.”
[chuckle]
0:22:03.8 DW: If the assumptions you make don’t break the symmetry of the problem, they don’t distinguish past and future in some sense, it doesn’t matter how clever your argument is. You will either fail to get out an asymmetry at the end of the argument, or you will have cheated somewhere and smuggled in the asymmetry, possibly without even noticing it yourself that you’ll have smuggled it in. Just as a matter of logic, you can’t have an input that doesn’t distinguish a direction of time and have an output that does distinguish it.
0:22:32.2 SC: So what do we do? What is the right answer? Where does the asymmetry of time come from, then?
0:22:38.3 DW: Okay, well you can think of three strategies, and one of them we can probably put aside, which is this idea that it comes from something non-physical, it comes from human intervention or something.
0:22:48.8 SC: Okay.
0:22:49.6 DW: So I think for our purposes we can put that aside and continue in our physicalist dogma. Within physics, if you’re going to break that symmetry, I think you can only logically, you can only really do it in two ways, your dynamics needs to violate the direction of time, needs to distinguish past from future, or your boundary conditions needs to distinguish past from future. So the former approach needs you to change deep physics, put that aside for the moment. That’s another conversation we’ll have, and it gets on to something I think we’ll talk about later, which is how quantum mechanics plugs into these questions.
0:23:31.7 DW: Assuming you don’t want to change the microscopic physics, you’ve got to say something about an appropriate initial condition. What often happens is what you end up having to do is say something that gets pushed earlier and earlier in time. So let’s suppose I do one of these really innocuous-sounding statements about, “It’s really probable that the gas will spread out. It’s really probable that entropy will increase, etcetera.” If I place that condition now, if I say, “It’s a condition on the world now, that we have really high probability entropies going out into the future,” or more plausibly, if I have some condition that looks really innocuous, it has that consequence.
0:24:12.0 SC: Yeah.
0:24:12.1 DW: But again, as a matter of time symmetry, it’s going to have the same consequence backwards. And the forward result is what we want, and the backward result is what we don’t want. You could try take it early, you could say, I’ll impose a bounded condition 100 years ago, which has the consequence that we would expect gases to expand and entropy to go up and so on. Now, the last 100 years is fine, and now everything we thought we remembered about everything prior to 1921 is completely wrong. We could put it back a million years ago, and that saves human history, but it’s too bad about the dinosaurs.
0:24:46.0 DW: We can put it back a billion years ago, in which case the dinosaur are fine, but it’s too bad about formation of the solar system. Eventually you end up saying, “I need to,” again, if we’re sticking with this sort of broad picture, we want to be reductionist, we don’t want to rely on external intervention, the only real option you’ve got left is to put some kind of relevant boundary condition at the beginning of time. So then you say, well, the universe, or at least the bit of the universe we’re aware of, the kind of observed cosmos, came into being, or very early on was such that it has these features, whatever they are, that’s another conversation, which established that entropy goes up into the future.
0:25:29.2 DW: And then you’ve kind of just… You’ve just kind of crossed out the earlier bits, so the paradox doesn’t occur. There’s a kind of sense in which you haven’t exactly broken the time reversal symmetry in physics jargon, the distinction between past and future, you’ve broken the time translation signature symmetry, the idea that there’s no preferred moment, in a sense you’re using the first moment as your preferred moment.
0:25:56.4 SC: I’ve always wondered… Again, an unfair historical question. Why didn’t Boltzmann or someone in his vicinity propose the big bang, why didn’t they take seriously… I mean, he did in a paper suggests that one way out of these reversibility problems was to have a special initial condition. But it’s really like a sentence in a paragraph and then he moved on with his life, right, and no one quite chased that down and said, “Look, you know, stars are burning some kind of fuel, we don’t know what it is, but presumably it can’t go on forever. There must have been a moment in the past when things were set up in a special condition.”
0:26:31.0 DW: Yeah, so historically I have no idea, I mean, trying to do a rational reconstruction… I think part of it is… Until you’ve got the general theory of relativity, any decision that a particular moment is the first moment is just a labeling choice. You can imagine saying, well, this is time equals zero, this is the first moment. It is forbidden to run the dynamics backwards from the first moment, but then your graduate student says, well, look, it’s a free country, I’m going to run the dynamics back anyway.” And then you’ve got a perfectly reasonable description backwards, then you’ve got effectively a symmetric cosmos that has a big bang in both directions, which is a defense…
0:27:13.7 DW: I gather some quite smart people have played with that idea. But it’s not big bang idea exactly. I think it’s when you’ve some… When you’ve got some idea of relativity on the table, you’ve got the idea that the theory might enforce the fact that a moment is first, or… As you know, not literally that a moment is first, but that there’s a point beyond which you can’t…
0:27:30.2 SC: There was some special moment in the history of the universe, right.
0:27:33.7 DW: I think that’s part of it, but I think also in a way I don’t entirely understand, there was clearly a really strong philosophical prejudice against the big bang. The advocates of steady state cosmologies did so quite a long time after the empirical case for it was looking shaky. They were clearly very philosophically motivated for it, beyond the data. As everybody who knows the history of this at all knows, Einstein could have predicted the big bang and he instead cooked up his equations to stop it happening. So there was clearly something in the air I don’t really understand that made people really skeptical about that kind of beginning of time concept.
0:28:09.2 SC: And this is… This idea that we need to set up the initial conditions in a certain special way. Number one, it truly blows my mind, even to this day, that the reason why ice cubes melt and don’t un-melt has something to do with what was going on 14 billion years ago at the big bang, and so I get a lot of mileage out of that, and I’m still impressed with myself. And number two, there’s this label for this idea, which is the past hypothesis, and then we argue about what about past hypothesis really means. So what is in your mind is… Why do we call it the past hypothesis and what is it?
0:28:42.4 DW: Okay, so the easy answer to why we call it this is because David Albert is good with technology and came up with its name, or I think… Or popularized, I’m not sure. It’s basically his language, I think. So that’s the shallow answer. It’s some… In the minimal version we’re just saying something about the initial universe. Some of the reason it gets talked about as a hypothesis is there’s a kind of tradition of thinking that somehow this assumption about the past is not supposed to be the kind of thing we might make in the same way we know other things about the past, somehow it’s not just like we… We put up a satellite and the satellite told us something about the early universe and so we’re using that as our boundary condition. It’s supposed to be something more… It’s a presupposition for the applicability of the physics we’ve been using, including the physics of satellites assumes that the very early universe has this feature.
0:29:38.2 DW: In terms of what that feature is, I mean, that’s contentious. You and I have been on different sides of several academic conversations about this. A very popular idea is, the condition is something like the very early universe has a very low entropy, is in a very special state in the probability senses I was using earlier. There’s another sense, which I prefer, which is actually something like, it’s more that the very early universe is in an appropriate sense, not in a special state, it’s in a state that doesn’t have the sort of very delicate, precise correlations between where all the particles are that would mean that you weren’t in one of these probable conditions whereby entropy increases.
0:30:22.3 DW: So just how to cash out this condition, it’s a pretty subtle matter. You can kind of work out just by logic that there has to be some condition of that kind. If you’re not breaking the symmetry on the dynamics, you’ve got to break it in the boundary conditions, but quite what the right way to break it is… Yeah, it… The devil is going to be a little bit in the details.
0:30:46.0 SC: Maybe this is actually… It’s always fun, but also dangerous to get too much into the weeds, but maybe there’s a weed or two that we can get into here, because it’s worth driving home, because even professional physicists struggle with this idea. One way of stating the issue is that, look, we have satellites up there, they’ve taken pictures of the early universe, right, the cosmic microwave background, it’s very smooth and almost homogeneous, but with tiny little ripples in it, and we explain that in terms of a smooth Big Bang 14 billion years ago, and let’s take for granted right now the idea that that condition of a hot, dense, smoother early universe is low entropy. We can talk about that too. Let’s say that is low entropy.
0:31:34.1 SC: So there’s an attitude that would say, look, we’ve taken a picture of it, it’s not a hypothesis, we have data that says it was low entropy. But then the much more subtle counter-argument is in the space of all possible ways that microwave photons could have come to our telescope and looked smooth, it turns out that most of them do not correspond to an early universe that actually is smooth, right. There are many, many more ways the early universe could have been wildly fluctuating, but with very very subtle coincidences and conspiracies that made the radiation get to us and look smooth, and it’s that kind of perverse behavior that you in the past hypothesis are trying to rule out. Is that fair?
0:32:15.4 DW: Right, I think that’s fair, although let me give a caveat, ’cause you were helping yourself a moment ago to get to this kind of, well, there’s vastly more ways it could be this than could be that. And I was using that kind of way of talking earlier, but actually, I think if we push on this, it starts getting us into trouble. I mean, I said sort of flippantly, well, there are infinitely many ways the system could be, so there’s no sense in which it could be more one way than another, but that is kind of true in a way, and even if there were like a million billion kazillion kajillion ways the system could be that it’s finite, it doesn’t follow from the fact there are a million zillion billion kajillion ways a system could be that it is equally likely to be each of those ways.
0:32:54.2 DW: I think we’ve got a ton of evidence from the actual application of physics that there’s some robust notion of probability in statistical mechanics, but I also think we’ve got quite a lot of reason to think that that notion of probability is not something we can get at a priori from just counting possibilities.
0:33:11.6 SC: Okay.
0:33:12.0 DW: There is something substantive additional going on. The assumption is roughly something like for a given macroscopic way the world could be, that it’s again, leaving aside my caveat about infinity, each of the microscopic ways it could be conditional with that macroscopic where it could be as equally like, that’s something like the way of stating it, but it’s quite a big step, that’s something I think there’s empirical evidence for, at least with some qualifiers and caveats and worries about the direction of time, but at some level there’s empirical evidence for it when we make that assumption and we derive equations as a result of that assumption, and we check the experiments in the lab or in the sky, the equations get the answer right.
0:33:51.9 DW: So that probability assumption is evidenced, we don’t need to think of it as a priori. The view that somehow if there’s lots of different macroscopic ways a system could be, then it’s more likely to be in the macroscopic state that has broad microscopic states corresponding to it. Well, there are special circumstances in which that’s evidenced as well, when systems have reached equilibrium, when they’ve had lots of time to interact and move around the space, then there’s evidence that’s true, but as a kind of a priori statement about the world, I don’t think it’s something we’ve got evidence for or a good logical reason that it has to be true. So while in a certain sense, you’re clearly right that there are many more ways, it’s much more probable that the universe fluctuated into the way it is now, rather than the satellites gave us the right answer, there’s another sense in which that’s non-innocent, that there are some assumptions about how to think about probability that are doing some work in that argument.
0:34:48.2 SC: I think that’s perfectly fair, if maybe a little bit persnickety, but that is your professional job, right, you’re a philosopher so that’s okay.
0:34:54.9 DW: Pretty much a job description, yeah.
0:34:56.3 SC: Exactly. We’re not going to hold that against you. So let me just rephrase it very quickly, just so we keep separate what the contentious parts are and what the parts everyone agrees with are. You’re probably right that since my job description is physicist and not philosopher, I speak sloppily, that’s my job description, right? [chuckle] So I use the word probability where I really meant number of different states or something like that, I do think that philosophers can sometimes be too persnickety about counting the number of different states. I think there’s a perfectly obvious way to count the number of different states, and that doesn’t mean that it’s a probability distribution, that doesn’t mean that in Penrose’s famous formulation that God threw a dart board randomly at that space of states, but there remains the fact that the past hypothesis needs to have some way of saying that in the space of all the different ways the universe could have been, there’s a measure, there’s a way of counting, and our real universe was in a very, very tiny region in that space, and that needs to… That’s the past hypothesis broadly construed.
0:35:56.5 DW: Yeah, okay. So I don’t think this is just Penrose. I think we actually disagree on something substantive here. I mean, there’s a well-known way of stating the past hypothesis that says something like you said there, that the hypothesis is that the early universe was in a very low entropy state or a certain particular state that is in fact very low entropy, and that if one assumes that, then one gets the direction of time right. I actually think that’s the wrong way to state the hypothesis, and one way of getting at that is, I think… This is like a minor bit of it.
0:36:33.6 DW: However low entropy the initial universe was in, it could have ended up in an even lower entropy state later. The issues about time reversibility are still there. Let’s suppose that the early… Okay, so firstly, actually, let’s say the early universe was very hot and very dense and we want… And we’ve established that we think that’s a low entropy state. It’s still a sense in which it could have been hotter and denser and it would be low, it would be hotter and it would be low entropy, so there’s a certain worry there, but more seriously, suppose that was a very low entropy state, there’s still a way in which it could have fluctuated to an even lower entropy state.
0:37:08.7 DW: So we don’t rule out that possibility just by specifying this very low entropy initial state, we have to say something else. And the something else, I think, takes the form, I actually helped myself to the physicist’s sloppiness, that given that initial state, it was equally likely to be in any of the microscopic states corresponding to it, and so it was super unlikely to be in one of the… Fluctuated to an even lower entropy state. But my feeling is, once you’ve said that, once you’ve said, okay, given the macro state of the early universe, it was equally likely to be in any of the micro states compatible with it. I don’t think anything more is gained, at least for the purpose of explaining the arrow of time in the here and now, by saying, and what’s more that state was a really low entropy state. If it was a really high entropy state, the world would not look like this.
0:38:01.3 SC: The world would have probably not looked like this…
0:38:03.0 DW: But it’s also true that if it was in a state of a different kind of matter-antimatter ratio it would also not look like this. The smoothness of the early universe explains lots of things about the here and now, but I’m not compelled by the argument that it explains the time directedness of our macroscopic physics.
0:38:20.1 SC: So let me see. So in other words, you’re saying… So actually, let’s back up a little bit. There’s the idea of a macro state, which we sort of glossed over, and people want to make a big deal about this, and I think it’s proper to make a big deal about this. So one of Boltzmann’s genius moves was to say, look, there’s atoms and molecules, they’re moving around, but I can’t see what the atoms and molecules are, so I’m going to group them, all the atoms and molecules that look similar, into a single group that I’ll call a macro state. I don’t know if he used those words, but that’s what we do. And so people have been asking, so I will ask you, what’s the license to do that? Is that okay, is that objective or is human agency sneaking its way in here, something subjective?
0:39:02.3 DW: Good. Yeah, I mean, the words get used in two kind of different ways, and I think one of them should worry reductionists, naturalists. So one of them basically shouldn’t, and the one that should worry people is the one that defines macroscopic in terms of our own abilities to distinguish them. So you want to say, well, the reason why we treat two different configurations of gas in the room as being the same macro state is something like, “We can’t distinguish between them with our eyes,” or sometimes even like, “We’re only interested in the gas at the level of a course graining,” like a macroscopic description.
0:39:42.0 DW: But #EmbarrassingConfessions, I don’t actually care about the gas at all. Beyond the minimal requirements of being able to breathe, I didn’t really mind where the air in my room is. It’s still true that thermodynamics describes it really well, even though I’m not curious about it. And similarly, we use statistical mechanics to describe even things like the way galaxies, stars in the galaxy develop, and we can obviously just… Our telescopes can tell us where the damn stars are, and yet we still want to dump various clearly visibly distinct arrangements of stars into the same macro state and say, “What we care about is the average distribution of the stars,” or something.
0:40:19.9 DW: So I think the good notion of macro state, the one that the naturalists can be relaxed about, is something like, “This is a level of description at which you can get some kind of robust, autonomous, in the philosophers’ jargon sense, self-contained dynamical story.” So, if I want to know how the air in the room is going to evolve over the next five minutes, at a certain coarse grained description, it turns out that all I need to know about the air now is, again, at a certain… Well, modulo these worries about thermodynamics it’s just that empirically, it’s true that I can make reliable predictions about what the air will do at a certain coarse level, provided I know what the air is doing now at a certain coarse level.
0:41:10.8 DW: And you know, we have equations to say what water does that don’t require us to say where all the individual molecules are. We can do higher-level science, and the macro states are the way of breaking up the lower-level stuff into higher-level stuff such that you can do higher-level science. That’s the robust version, I think, that we shouldn’t be suspicious about.
0:41:31.0 SC: Yeah, and I think…
0:41:31.3 DW: And I… Go ahead, sorry.
0:41:33.0 SC: I was just going to say, that feeds into not just statistical mechanics, but bigger issues of emergence. And Dan Dennett, a former Mindscape guest, has this wonderful phrase, “real patterns,” that you make use of in the case of the Everett interpretation. The fact that you don’t need to know absolutely everything about the universe but you can have incomplete information and still, in the right circumstances, make predictions about it is an objective fact about the universe, and that’s what we appeal to in this sort of Boltzmannian coarse-graining.
0:42:05.9 SC: Yeah, absolutely. And it’s as solid a fact that reductionism itself… I mean, it’s just empirically true that we can write down reliable equations for what gases do.
0:42:13.7 SC: It didn’t have to be that way, no.
0:42:15.8 DW: Now, if you can… Often we found those equations ’cause we did microphysics and worked it out, but not always. And even if we’d done the microphysics and worked it out, we could’ve thrown it away and still checked the equations and they were still right. So, insofar as your best theory of microphysics says it’s not possible to have equations of that form, so much the worse for your best theory of micro-physics. Our actual evidence for our microscopic physics mostly doesn’t come from microscopic measurements. Mostly it comes from these millions of intermediate calculations. So yeah, we talk about the standard model explaining everything or underpinning everything. There’s almost no experiment that counts as a systematic test of the standard model as a whole.
0:43:00.1 SC: Sure.
0:43:00.7 DW: As you well know, there are approximations and chains of approximations of approximations that ground all the various ways these theories work. And the first order of evidence is the evidence that these various approximations are right, and because they’re right, then that gives some support for the theory from which we derive them, but really the world we find empirically in the first order is a world just full of all of these different processes at different levels that we think are connected, and I think we’ve got good reason to think are connected, but which we have evidence for even if they’re not connected.
0:43:31.7 SC: Good. And specifically, so there is some sort of objectiveness to the coarse-graining that we choose in practice, and specifically because some of my favorite super smart cosmologists don’t understand this, the early universe has a low entropy in that coarse-graining. There’s this feeling that you say something like, “Look, it’s just a black body radiation, that’s high entropy. How could it… It had the highest entropy it could have had, given that it was tiny,” but when you turn on gravity, that’s not true. So am I right to say that it’s fair to say the early universe had low entropy compared to what it could have had?
0:44:10.6 DW: Yeah.
0:44:11.2 SC: And gravity had something to do with that.
0:44:12.7 DW: But look, here’s something I don’t understand about this, these conversations with cosmologists. Absolutely, the early universe had low entropy compared to what it could have, but one really quick way to tell that is the early universe had low entropy compared to the current universe. If it didn’t, the second law of thermodynamics is empirically wrong. And no one thinks that.
0:44:34.1 SC: Well, yes, but the…
0:44:35.4 DW: So it’d really better be the case that the early universe was lower entropy than the present universe. We shouldn’t need to give some subtle transcendental argument as to why it must be lower. It’d better be lower in order not to blow our entire thermodynamic theory out of the water.
0:44:48.3 SC: But to be fair, the argument, which is completely wrong, but still, it’s not completely nonsensical, is that there’s sort of an envelope of maximum possible entropy just because the universe was smaller, and the argument is that it was as high entropy as it could have been, even though it’s lower entropy than now. But even that argument is totally wrong because it ignores gravity and black holes and things like that.
0:45:13.3 DW: Yeah, it does, but I’d also say that it actually ignores the fact the universe is expanding, and I think even that homogenous piece of it, in some ways, like smooth, is enough to do the work we need of it here. The very early universe was in this equilibrium thermal state, then it expanded and that was the highest entropy it could be given the other constraints on it, and given that it was uniform.
0:45:39.9 SC: Given that it was uniform is a huge constraint.
0:45:42.6 DW: Absolutely, but look, go forward a little while, and it’s still uniform, but it’s no longer the highest entropy it could be. There comes a point where all the quiet gluon slush has turned into neutrons and protons. And there comes a further point when some of the neutrons and protons are fused into helium. There comes a point when it’s about three-quarters hydrogen and one-quarter helium. And then it stops there, and it carries on being three-quarters hydrogen and one-quarter helium. If it had wanted to stay at equilibrium, it wouldn’t have done that.
0:46:15.0 DW: The equilibrium condition for the universe… This is more your territory than mine, but well before it gets non-homogenous at large scales, the equilibrium configuration would have been all helium, and it wasn’t all helium. Why? Well, it was expanding so fast that the particles got too far apart before they finished crashing into each other enough to make helium. I’m going to just use that as an illustration. Well, you’re absolutely right, of course, about the fact that gravity and things not being uniform mean that the real story about entropy in gravitating systems is just radically different from non-gravitating systems.
0:46:58.4 DW: But even at a more basic level than that, trying to apply equilibrium arguments to a system that is very rapidly expanding is itself really dubious. Really rapidly expanding systems are not at equilibrium. And the universe’s macro conditions were changing very fast, so it was only ever reasonable to suppose it was at some kind of local equilibrium. You can think about the very early universe while it’s still uniform as being sort of like a piston. If I put some very hot gas in a piston and I expanded it very gently, then the gas would always stay instantaneously at equilibrium. And if your piston was ridiculously strong, you could compress it, you could expand it from being a quiet gluon plasma all the way to being cold iron, and it would all be kept at equilibrium.
0:47:39.9 DW: If you pull the piston apart a bit faster than that, then it’ll freeze out at some mixture of stuff that hasn’t finished going to equilibrium. It did that because it was expanding, it was expanding too quickly. And so, I think those things are going on in the early universe, even before you start considering gravity and non-uniform distributions. And when you think that the Earth is powered by fusion from the Sun and the fusion from the Sun is possible because there was lots of hydrogen left over from the Big Bang, which wouldn’t have been left over if the universe had stayed at equilibrium as long as it was uniform, then you realize that the thermodynamics of the present day actually rests on some of these factors as much as it rests on issues of non-uniform systems of higher entropy.
0:48:29.8 SC: I’m tempted to go back into the the weeds on the past hypothesis definition, but I worry that we’re getting too much into the weeds on that. There’s a lot of non-weeds things that I want to get to. [chuckle]
0:48:40.8 DW: Well, I think there’s a rather unfair thing I’m doing here, which is, this is actually a place where we have some minor but quite interesting intellectual differences, and I’m pushing my side against your position as a podcast host, who can’t fire back with a full force.
0:48:52.1 SC: You totally are, and I’m not even sure whether I disagree with you or not, so we should have an offline discussion and then we’ll put it somewhere on the internet to reward. We’ll make it as a reward for the Patreon subscribers or something like that.
[chuckle]
0:49:05.5 SC: Okay, but you said something that I think is very important to this bigger picture discussion. As the universe expands, there’s this competition between things trying to equilibrate, photons and electrons and protons and neutrons and helium, versus the fact that the universe is expanding. And so, it doesn’t stay in equilibrium, because otherwise, like you said, the hydrogen would just keep fusing, the helium would keep fusing, you would create all the elements up to iron and then you would stop there. So, the hidden fact that is relevant there is that even though the second law says entropy increases, there’s no law that says entropy increases as fast as possible. Sometimes people try to invoke laws like that in living systems or dissipative systems or something like that. Just very, very big picture, how far do you think we can go beyond the second law to be quantitative about the ways in which entropy increases?
0:50:01.8 DW: Yeah, so I’m not sure. And I’m not even sure entropy is always the reliable language to use here. One of the things I tend to do as a philosopher of physics is try sometimes to step down a little bit from the big picture of how this could possibly work and say, well, what are the ways in which it’s actually being done? And what are the assumptions that are going into them? We’ve got a very developed theory of non-equilibrium statistical mechanics, in the jargon, of the physics of systems that aren’t at equilibrium and are genuinely changing at time, but are big enough that you need to use these kind of averaging methods of statistical mechanics and thermodynamics to study them.
0:50:38.7 DW: And what you find in those stories is that entropy basically seems to be a book-keeping device. You can tell reliable stories about irreversibility, and normally, those are processes in which the entropy of the system’s going up. And when they’re not, it’s only because there’s some environment that’s absorbing the entropy cost. But the dynamic… Well, in the language I was using earlier about robust, autonomous structures, we don’t there seem to have a robust autonomous dynamics of entropy. You need to know more about the system than just its entropy to know what it’s going to do. And different systems do radically different things. And there doesn’t seem to be any general law that systems want to really go to equilibrium at all.
0:51:20.1 DW: Here’s an example that, at an abstract level, has just the same structure as the hydrogen-helium example, but is perhaps easier for people to grasp. If I take a box and I fill it with a mixture of hydrogen gas and oxygen gas, I don’t know, or a glass jar or something, and I leave that on my shelf, it can sit there for years and be perfectly happy sitting there. And if you think equilibrium is the state that things make their way to fairly quickly and then stay there, you might be forgiven for thinking that that box of a mixture of hydrogen and oxygen is at equilibrium. But if you strike a match…
[chuckle]
0:51:52.3 SC: You’ll find out.
0:51:52.7 DW: You’ll really quickly realize it was not at equilibrium and it’s much more entropically-favorable, much more thermodynamically-favorable for the hydrogen and oxygen to react and form water and blow your lab up. But there just happens to be a dynamical bar towards getting there, and it’s not… It’s not a mysterious bar, it’s about the fact that the typical energies of… There’s a certain amount of energy of which hydrogen and oxygen need to crash together in order to turn into water, and that typical energy is much higher than the typical energy of actual hydrogen and oxygen in the gas. And so typically they just bounce off each other and keep going.
0:52:27.7 DW: If you waited really, really long times, it would happen, but it’s just not going to happen spontaneously in the time scale when you’re sitting in the room. So on some level we understand the physics of that fine, but the point is it’s that specifics of the physics that tells us that this system does not have increasing entropy. I take from examples of that kind that there can’t be uniform or universal principles about how quickly entropy increases of something. That’s not to say that there might not be powerful generalizations that are not universal, but nonetheless have a universality in physics jargon, apply to a very large class of systems.
0:53:07.1 DW: As far as I can see, we don’t very systematically have things of that kind, so far, but that doesn’t tell us anything about whether we could have things of that kind. I don’t know, this is more that kind of, what are the most promising routes to develop an exciting new piece of physics, is a place where I, to some extent, am now outside my comfort zone, so I’m not sure.
0:53:32.5 SC: But I think that…
0:53:33.8 DW: Yep.
0:53:34.1 SC: It’s an excellent point to make that, okay, there’s the second law and entropy goes up, but that doesn’t mean that entropy is necessarily the best thing to be keeping track of when you’re trying to formulate these rules. And in fact, I would spin it in a positive way as saying, as a researcher, there’s a lot of landscape out there for potential discoveries of better ways to state these universal rules that we don’t have. Non-equilibrium changes in statistical mechanics is a rapidly growing field right now, and sort of has been sadly neglected in the previous centuries, so I’m optimistic about that.
0:54:08.3 DW: Hmm.
0:54:09.8 SC: Well, good, so this is a good sort of segue then into these, remaking the connections we started with, to other arrows of time, so we have a complicated universe full of all sorts of different things with manifest… Manifestations, I should say, of time asymmetry, which we’re going to say are ultimately due to, let’s call it the thermodynamic arrow of time, the fact that entropy is increasing, however you want to characterize that arrow of time. Why does the fact that ice cubes melt and don’t un-melt help me understand why I remember the past, but not the future, or why I get older as time goes on?
0:54:46.0 DW: Yeah. I’m not sure I know the right general thing to say here, and I think a lot of the attempts to say it have never been, I’ve never found completely compelling. Here’s a starting point that I think matters, when we talk about the way in which the physics distinguishes the past from the future, sometimes we characterize that as just there isn’t a symmetry that reflects past and future. But that, I think, doesn’t fully get at it. I mean, there are, as you know, there are microscopic bit of physics that in a certain sense genuinely break the symmetry of past and future, the decay of neutrons in a certain way, or certain… Rather more esoteric than that, I guess, actually, for that symmetry. Esoteric bits of particle physics actually seem to break the past and future symmetry, but not in the way that matters from this point of view.
0:55:39.6 SC: Right.
0:55:39.6 DW: Not in my, if you ran the video backwards, would it be manifestly obvious that you were doing anyway. What I think matters in the way these dynamics run differently, forward and backwards, is they have irreversibility and they have noise. Irreversibility, meaning you lose information, lots of different initial states come together and end up in the same final state. Noise in the sense that it becomes, at least de facto, impossible to predict which of many future states the system is going to make its way into. And those features of irreversibility and noise wash out information in the evolution process. You could see this if you look at predicting forward and backwards.
0:56:30.0 DW: If I’ve got something like the solar system physics, like I was saying for eclipses, the way I go about predicting what the solar system will like in a thousand years time, it’s exactly the same as the way in which we predict the… Well, retrodict is the jargon, the way the world will be, was a thousand years ago. And we say this funny neologism retrodict in this situation, because we really are just doing a prediction, just with time set as negative. Mostly, you can’t retrodict macroscopic laws. If I’ve got the ice cubes in isolation or I’ve got my… Let’s say that I’ve got my cup with a little bit of ice and there’s cold water around it, I can predict pretty reliably the rate at which that ice cube will melt and the length of time at which the ice cube has gone entirely.
0:57:14.6 DW: I can’t retrodict that way, I can’t work out which ice cubes were there before, how they were put in, that information is just lost. Maybe the microphysics still knows it, but at the level of description we’re talking about, it’s just gone. So I think at some level, these kind of ways in which our dynamics are irreversible information losing, are also in the business of distributing stuff across space in a way and not bringing them back in. These are the kind of features of dynamics that then suggest that if you try to build things in that dynamics, if you try to do information processing in that dynamics, that information processing is also going to have to have a clear distinction in the way it treats the past and treats the future. So that’s the proof of concept.
0:58:06.0 DW: I don’t think I’ve ever worked out fully to my satisfaction how the details of that are going to go. Why does that general asymmetry of information processing specifically lead to the fact that we’re putting down records about the past, and making partially reliable predictions about the future and not vice versa. I think the quick attempts to do it are not convincing. The convincing attempts aren’t fully there.
0:58:30.1 SC: I mean, there seems to be at least a naive tension between what you said, which is certainly true about the fact that there’s dissipation and we lose information in some real sense with the fact that as a real person, I think that I’m learning things as time goes on and I’m writing books and there’s information in the books, and so there’s some local increase in the amount that I know about the past, and that’s clearly going to be a little bit tricky to reconcile.
0:58:53.8 DW: Yeah, and look, I guess you know, of course, by the same token, there is an apparent tension, it’s long been recognized, some people who should have known better have used it to argue against science or for all sorts of misuses. It’s long been the case that there’s an apparent tension between the fact that entropy always seems to go up and the fact that we go towards equilibrium and the fact that I personally am not at equilibrium and don’t expect to be at equilibrium for some decades.
0:59:15.8 SC: Right.
0:59:17.8 DW: And of course, the answer that is unmysterious, I am not a closed system. I can maintain my own low entropy by dumping entropy into the world around me, just like my fridge can stay cold by warming up the room, and at some level these kind of information processing things have the same answer, I think, you can… Even if the physical processes that are going on are systematically in some senses losing information, then I… Provided I don’t mind dumping the world into randomized states, I can still maintain an increasing information about the things that interest me.
0:59:58.4 SC: Which brings up a related worry that people often try to pin point me with. If you think that all of the differences between past and future ultimately come from this thermodynamic gradient, if someone puts you in a refrigerator and your entropy starts going down, do you suddenly remember the future, but not the past? I mean, is it possible to sort of completely reverse the arrow of time locally in a way that would make physics seem very weird?
1:00:25.1 DW: Yeah, well, not by putting you in the refrigerator. [chuckle] It kind of goes a bit to what we were saying earlier about entropy not necessarily being the sole measure of irreversibility. I mean, there are a ton of irreversible processes in your body, and even if you’re cooling down so that your overall entropy is decreasing, those processes are still happening irreversibly. The dynamics that’s governing all of the kind of biology in your cells is still happening in a way that draws a clear past/future distinction. Now, if you did something to you that systematically reversed those directions, that’s another matter, but now you won’t do that by a fridge, now we need kind of space aliens and God-like science to do it, and we kind of know what would happen if you were run backwards and if your environment was preserved.
1:01:10.6 DW: So if your environment was preserved to be compatible with it, then obviously it would just be like… Leaving aside, I think, a slightly badly phrased question about what that would be like subjectively, then, from the point of view of the person watching you would just do everything backwards.
1:01:24.9 SC: Right.
1:01:25.4 DW: But that’s the… Obviously, we set the problem out that way, that’s trivial. If your environment was not also running backwards, I think extremely quickly the environment would mess up… Would mess with the way you were trying to run backwards and start you running forwards again.
1:01:41.9 SC: Yeah, so this crude kind of thing, like just lowering your entropy overall is not nearly enough to truly reverse your personal arrow of time.
1:01:50.9 DW: Exactly. All of… If what we’re really after here is dynamical irreversibility, then the dynamically reversible processes are not changed by… In effect, what’s going on in entropy terms, all of those processes are still generating their entropy, and we’re also thwacking a big chunk of entropy out of your system just by moving bits of your body that are locally at equilibrium to bits of your body that are at local equilibrium at a lower temperature, and that overall entropy budget, which is a sum of the various non-equilibrium bits of your body doing non-equilibrium stuff that increases entropy, plus the non… The ordinary cooling down bits of you, all of that is going to be… Going to be negative so that your entropy drops, but it’s still the case the dynamics in your body is clearly distinguishable between past and future.
1:02:41.6 DW: It’s just that that crude indicator of which way is the entropy gradient going, isn’t doing it for you. There’s more to thermodynamic irreversibility than just the entropy gradient.
1:02:49.6 SC: Well, I believe everything you just said, but I think you’re missing a big opportunity for a start-up that would not only preserve you in time, but actually make you younger just by putting you in a refrigerator. If you could develop some theory of that, you could be making beaucoup bucks, I’m just saying.
1:03:05.6 DW: If it worked, I think I can make a lot of money. If I could convince people it would work, I could be making almost as much money, but I’m a little bit too honest.
[laughter]
1:03:12.2 SC: The convincing people work is an easier task than making it work.
1:03:14.7 DW: Oh, yeah.
1:03:14.7 SC: So, I think that’s the whole secret to being in this business, but, okay, alright, well, good. With all that wonderful stuff on the table, there is a looming specter over this whole discussion, which is that we’ve been speaking pretty classically, right?
1:03:30.1 DW: Yeah.
1:03:30.3 SC: The good old… We even mentioned Isaac Newton and Maxwell and people like that. And there is a famous part of physics where it does not seem manifestly time symmetric, which is the active observation and the collapse of the wave function in quantum mechanics. So you know something about quantum mechanics. How do you think about either generally, or you’re very happy to dive right into Everett and many worlds if you want, but how do you think about the time asymmetry of measurement in quantum theory?
1:04:00.0 DW: Good, okay. So my starting point is going to be that same kind of dynamics first, broadly reductionist starting point that I think we just sort of took as a common starting point an hour ago. And that’s going to be… So, measurement here is just a special case of large-scale interaction. Measurements happen when some relatively small-scale thing gets magnified up in a highly redundant way. The processes of measurement from that point of view are going to have to be special cases of this process of being magnified to the macroscopic, but of course, famously in quantum mechanics, that process is problematic.
1:04:40.0 DW: The process of measurement of macroscopic amplification de facto looks something like the randomized collapse of the wave function. As a practical matter, it looks as if what was originally a superposition of lots of possible goings-on on snaps into being one of them. What’s really going on to justify that is another matter, but that’s kind of how it just looks from the point of view of doing the practical physics. The emergent level description is a description where I move from a superposition to a randomly selected element of the superposition. That process is blatantly, manifestly time irreversible.
1:05:18.5 DW: And it’s always amazed me, actually it’s less true than it used to be, but for a long time it’s always those amazed me that that kind of quantum mechanical arrow of time was the sort of, I don’t know, the forgotten stepchild of the arrows of time in physics, it got very little airtime compared to the sexier arrows of time, if you like. But really it’s… Because we say quantum mechanics is kind of fundamental, it really ought to kind of matter here that it’s turning up, and I think the reason why it doesn’t… Well, there’s a claimed reason why people don’t worry about these quantum effects and there’s a real reason, I think.
1:05:58.0 DW: The claimed reason is something like there’ll be a standard line you’ll get at the beginning of almost any sort of textbook on philosophy of statistical mechanics. It’ll say something like, “Yes, I’m using classical mechanics. Yes, the world is really quantum. Yes, we’d better allow for the quantum effect, but quantum mechanics is technically… Quantum statistical mechanics is technically more complicated than classical statistical mechanics and essentially, the same foundational issues arise.
1:06:20.6 SC: I’ve said those words. [chuckle]
1:06:22.7 DW: That is, I think, just almost exactly wrong as a statement of what’s going on. I think actually for foundational purposes, I won’t go into the detail, ’cause it’s a bit into the weeds, the foundational process of quantum mechanics is easier. There’s a bunch of pathological features of classical mechanics that are not present in quantum mechanics, if you want to calculate that, Sean, heaven help you, but if you want to just work out the abstract structure of the problem, it is easier to be quantum. And the issues change profoundly, for a bunch of reasons, and I’m sure we’ll come on to more of them, but in the first instance, just precisely because of this clear asymmetry of the measurement of macroscopic amplification.
1:07:04.8 DW: I think the real reason people who write textbooks on this stuff or monographs on this stuff do classical statistical mechanics is they’re scared of the quantum measurement problem. And they don’t want their beautiful thesis about statistical mechanics to be adulterated by all of this quantum mechanical stuff.
1:07:18.9 SC: I would agree with that, if we interpret the word scared as having a concern that it would derail the entire discussion.
1:07:26.6 DW: That’s fair, yeah, and I think the problem is, it does derail the entire discussion, but that’s another way of saying that quantum mechanics is really important in understanding the statistical mechanical arrow of time. But I suppose put it this way, look at that asymmetry we’re talking about. That asymmetry cannot cleanly be distinguished from the statistical mechanical asymmetry, it’s not exactly the same as it, but it’s intimately related to it. If you… There’s a bunch of different ways of seeing it, but if you think that, for instance, almost every… If you think that the collapse of the wave function is a genuinely random process at a really deep level, for instance, it more or less, it almost destroys any concerns that I can run my probability argument just as well forward and backwards.
1:08:10.7 DW: If it’s extremely probable that the current state will evolve into a high entropy state, but there are certain super special ways that the current state could be… That could have evolved into a low entropy state, well, even if you put the system on these super special states, it’s not going to stay there for long, because quantum randomness will just kick it around a bit. So if you had fundamental collapse of the wave function, you would break the asymmetry in time of the dynamics in a way that it’s heuristically really plausible, would sort it out with statistical mechanics.
1:08:39.2 DW: Now, if I thought collapse of the wave function was a good idea, then I could go on at some length on how we could reform statistical mechanics. I think, for reasons that can be shared, that this is a terrible idea… Well, it’s a good idea, it’s a fundamentally wrong idea.
1:08:51.2 SC: It didn’t work.
1:08:51.2 DW: It’s worth exploring, but it doesn’t work, it’s not the right way to go. The right way to go, I think, is something like the many worlds interpretation. We want to say that quantum mechanics really maintains all those superpositions, and the fact that it looks as if we’ve got one of the possible outcomes at all there at once, rather than all of them at the same time, is perspectival. It’s that we just happen to be the kind of physical system that’s sitting in one of those terms, there’s a David in one of the other branches after a quantum measurement who’s just as committed to being David as I’m committed.
1:09:24.2 DW: And of course, there’s a whole panoply of issues associated to many worlds quantum mechanics anyway, but to stick with the statistical mechanical issues, many worlds quantum mechanics is time-inverse invariant, because it’s just unitary Schrödinger equation quantum mechanics, it’s reversible in the previous senses, and so there isn’t any prospect in many worlds quantum mechanics of having some prior quantum arrow of time that underpins the statistical mechanical arrow of time.
1:09:55.0 DW: In a sense, it’s almost the other one round, if you look at the kind of mathematics that’s used to describe measurement and the amplification of superposition to the macroscopic, and the thing that quantum physicists called decoherence, where it becomes impossible in practice to detect the many worlds nature of systems, because they get so tangled up with their environment. If you look at the actual mathematics of all of that, it’s all just statistical mechanical. The kind of assumptions made to derive those equations are just the same assumptions as we use to derive the equation in non-equilibrium statistical mechanics. In fact, sometimes they’re literally the same equations, there is a thing that people in decoherence theory and foundations of quantum mechanics use, the decoherence master equation…
1:10:43.2 SC: Master equation. Yeah.
1:10:43.4 DW: Which describes standard models of how the environment influences the system. It is literally the same as a standard equation that’s used in non-equilibrium statistical mechanics and has been used from the ’60s and ’70s. There’s nothing untoward there, no one’s denying that, but it’s surprisingly often not recognized, because the people working on one side of that divide aren’t super familiar with work on the other side either. I met that equation as a grad student interested in quantum mechanics in the ’90s in the decoherence literature. It was 20 years before I rediscovered it in non-equilibrium statistical mechanics. And okay, I’m a philosopher and I get to do other stuff, but even so.
1:11:20.7 SC: Well, yeah, I think we can take on board the idea that our listeners know that there’s something called quantum mechanics, and there’s a wave function as a superposition of different things, and even that they know that you and I are both fans of the many worlds interpretation which says that, like you say, collapses aren’t real, there’s just this smooth reversible evolution of the wave function. Okay, but then there’s branching, and I think that that is exactly where the time asymmetry comes in.
1:11:51.1 DW: Yes.
1:11:51.3 SC: But maybe before the time asymmetry, we should ask, because this is a crucial question, and I try my best to give good answers to it, but it doesn’t always satisfy people. What do you mean by branching? When does that happen, how often does it happen? How should I think about that?
1:12:06.7 DW: Good, good. Okay, so let’s back up a little bit. Here’s what I think about it. If people listen carefully, I’m going to smuggle in a whole bunch of assumptions about the direction of time that we’ll need to come back to, but ignoring that for the moment… So what’s going on when we measure a system or we allow a microscopic system to magnify up to the macroscopic is effectively that lots and lots and lots of other bits of the world get tangled up, get quantum entangled with the system in question.
1:12:39.6 SC: Entangled. Yeah.
1:12:40.1 DW: So the kind of thought experiment I always start with, I’ve got something like Schrödinger’s cat and Schrödinger’s cat is alive and dead at the same time, and the normal form of the quantum measurement problem says that that can’t be the way things are, because when I look at cats, I never see them alive and dead at the same time, and somewhere in the background there is the idea that what it would be like to see a cat alive and dead at the same time. It’s kind of like what it was like in the before days before the pandemic, where you just got really drunk with friends, and you were so drunk, you saw double.
1:13:12.0 SC: Yeah.
1:13:12.7 DW: And so seeing a cat alive and dead at the same time would be like having an experience of seeing a really weird indefinite cat. That’s intuitively what it would be like to see a cat that is alive and dead at the same time. Intuition is a really bad way to work out how things look in a physical theory. If you ask the physics what it would look like, well, if I look at a cat that’s alive, that’s just a physical process. So my brain goes into some state or my body goes into some state it might describe as a seeing a live cat state, and if I see a dead cat then I go into a state that you might describe as seeing a dead cat state. So what it’s… This Schrödinger equation is linear. So what that means is if I — don’t mind details of that — it means if I look at cat that is alive and dead at the same time, then my state after that is seeing a live cat and seeing a dead cat at the same time.
1:14:00.3 DW: I’m not myself in some definite state of seeing a weird indefinite cat, I’m myself in an indefinite state between two rather mundane cat-seeing experiences, and the cat’s connected up so the… Me and the cat together are jointly in a state of cat alive I see live cat, plus cat dead I see dead cat. And if I then tell you on the podcast whether the cat’s alive or not, then you end up in a being told it’s alive and being told it’s dead state at the same time and your listeners when it’s played… I mean, much sooner than this, really, but in fantasy for the moment, your listeners when it’s played are in a being told by Sean the cat’s alive plus being told by Sean the cat’s dead state at the same time.
1:14:41.6 DW: And again, it’s all correlated, so the whole quantum state of me and you and the cat and your listeners is a sum of two things, both of which are ordinary, one of them is the cat’s alive, I told you that the cat’s alive and you broadcast on the internet, and the other one is the cat’s dead, I told you the cat’s dead, and you broadcast that on the internet, and the rest of the world, as they hear about and find out and get entangled. Well, the reality, actually, is, it’s faster and less picturesque than that.
[laughter]
1:15:08.6 DW: The ordinary kind of electromagnetic disruptions and things for me doing this, will spread out at the speed of light and put you in a state of at the same time being correlated with David seeing a live cat and being correlated with David seeing a dead cat even before you’re consciously aware of the answer to that. So the Earth as a whole, the solar system as a whole pretty soon, is at the same time, cat alive and cat alive goes on and cat dead cat dead goes on, and each of those terms separately is not interacting with the other one in any meaningful sense. The way in which quantum… This quantum superposition, this quantum two things at once, show up as two things at once, is through interference effects that we can do between the two, which… Whereby somehow things can reinforce or cancel out, so somehow the live cat bits could cancel out and the [1:16:00.6] ____ could reinforce… That can’t happen here.
1:16:03.7 DW: Once systems get big enough, this is more thermodynamic irreversibility, actually, but once systems get big enough, you just can’t do those kind of interference, there are too many moving parts to mesh up like that. So what you’ve really got is two chunks of physical reality, each one is not interacting with the other one, and each one looks like, structurally it’s isomorphic to, has the same shape as what things would be like if an ordinary, definite classical thing is going on. So the world is two ordinary classical definite things going on at the same time, or the universe is, and we need a term for some big chunk of reality that’s doing its own thing and not really interacting with other big chunks of reality.
1:16:48.0 DW: World and branch are quite good terms for that, but they’re emergent. I mean, my book on this is The Emergent Multiverse, it’s not the sort of fundamental of the… At the deepest level of reality multiverse. All of this kind of multiple things happening at the same time, is just more of that higher level autonomous goings on that we were talking about earlier in the podcast, it’s just there’s more than one bit of it at a time.
1:17:08.2 SC: Right. And actually it’s worth saying for potential skeptics in the audience, if you signed on to the previous discussion about these emergent structures and how they could have autonomous laws, then you’re most of the way to signing on to the many worlds interpretation of quantum mechanics, it’s a very similar thing.
1:17:24.5 DW: Absolutely. This was one of my original arguments for many worlds in philosophy, like way back when I was making that jump from being a physics graduate student that once you accepted the basic rules of how emergents worked in physics and you applied them to unitary unmodifying quantum mechanics, the many worlds just more or less happened, they weren’t something you put into the theory, they were just given to you structurally by the way it was set up, and so conversely, if you, if you want to reject the Everett interpretation, you’re in some danger of rejecting rather general rules about how theories at different levels are related, rules that we use pretty much all over science, rules such that if we didn’t use them a lot of science would just stop working.
1:18:08.1 SC: But now I get to ask you all these tricky annoying questions, ’cause you helped yourself to a very simple example of Schrödinger’s cat which. By the way, fans of the Mindscape Podcast can buy a t-shirt with Schrödinger’s cat on it at the Mindscape merch store. Although my cat is alive, it’s awake and asleep versus a live and a dead, ’cause I don’t like to kill the cat.
1:18:30.3 DW: Well, you have cats, don’t you?
1:18:31.4 DW: Exactly, yeah. There’s room for cat people. But when the observational outcome is not something that only has two possible outcomes, but is continuous, like the position of a particle, is branching happening all the time? And can we even count the number of branches?
1:18:50.5 DW: Yeah. I don’t think we can count them, or at least we can’t count them non-arbitrarily. Look, the cat is an example of this, really. I’m going to stick with the brutal version of the original Schrödinger’s cat, apologies to cat lovers. There are many ways in which that cyanide capsule can break. And come to that, there are many times in which the event can happen. The way people tend to tell Schrödinger’s cat in physics classes these days is, you make a measurement of some particle and according to the outcome, the cat lives or dies. It’s very clean, it’s very discrete. But Schrödinger’s original one had a radio isotope, so we did actually get a discrete breaking into live cat and dead cat. We’ve got a live cat plus a continuum worth of dying cats who died different times.
1:19:38.0 DW: So, how do you analyze that in the kind of emergent story I was telling? Well, you can’t literally treat it as a continuum. If you want to say that I want to separate the various dead cat states into states that differ by, I don’t know, 10 to the minus 25 seconds. That won’t work. On those timescales, the quantum coherence of the radioactive decay matters. The degree to which the wavelength of the decayed particles requires constructive interference to make sense of it means if you try to break the world up into worlds that are that finely-grained, then my claim earlier on that there’s no interaction between these worlds, no interference effect between them, stops being true.
1:20:29.5 DW: At that grain of analysis, there’s interference. So, I can’t think of the cat state as being decomposed into branches that differ by 10 to the minus 25 seconds. I slightly pulled that number out of the air. I wouldn’t swear to the physics…
1:20:41.8 SC: Seems plausible, though, yeah.
1:20:42.8 DW: There’s a timescale such that that’s true, let’s call it 10 to the minus 25 seconds. If I decide to do it on 10 to the minus 6 seconds, it’s fine. The relevant coherence times for radioactive decay are much shorter than that, so I can definitely regard the cat state as a superposition of hundreds of millions of dead cat branches differing by 10 to the minus 6 seconds. Can I get it down to 10 to the minus 12 seconds? Yeah, probably. Can I get it to 10 to the minus 18 seconds? I’m not sure. And there’s no exactly sharp line where that stops working. There’ll come a point when interference becomes such that the description in terms of branches becomes un-useful.
1:21:32.5 DW: That’s the point at which there won’t be an autonomous high-level dynamics of that description. And that’s the point at which the real pattern stuff Dan Dennett talks about, that you were talking about earlier, that’s sort of my general license to say, “I’ve got higher-level goings on.” That license is revoked at the point at which you get to that finer grain. But on a coarser grain than that, everything’s fine. And there’ll be intermediate points where you can get away with it, but it’s a bit fuzzier than you might like. And sort of depending on your tolerance for error and the degree of the precision in which you care, you’re going to reach different conclusions as to exactly how finely you want to break things up. But the basic claim that there’s a vast number of different cats is robust against those various ways of precisifying it.
1:22:14.3 SC: And there’s a lesson or a philosophy implicit here, maybe you made it explicit. But look, these branches are just approximations, they’re useful ways of talking about things, and you can’t demand too much precision of them, just like you can’t demand too much precision of treating the gas in the boxes of fluid, rather than a bunch of atoms.
1:22:36.2 DW: That’s exactly right. This is, again, this general lesson about emergence being brought into quantum mechanics. It’s just systematically true that emergent accounts have fuzz and slightly blowy boundaries, and no really sharp, exact rules as to where you divide them out. That’s nothing to do with quantum mechanics, that’s just generally how the world is structured. There’s a temptation to talk about quantum mechanics and talk about many worlds in particular. It’s less now, perhaps, than it was, but it’s still around to say that there’s two options. Either something needs to be exactly and precisely defined in my microscopic physics or it needs to be an illusion.
1:23:18.8 DW: But hardly anything in the world fits those categories. You and I are not exactly and precisely defined. There was no Sean-David interaction term in the Lagrangian of the standard model. And yet, we’re not illusory, or at least if we are, then the word illusion loses its value here. Everything higher-level has this fuzziness to it. And again, the many worlds branches just have that fuzziness in the same kind of way. But the things that are still true no matter how much you tweak the set-up, those are robust things. So, it’s robustly true that Schrödinger’s cat splits into a vast number of branches. It is not robustly true that the exact number of those branches are 10 to the power of 25 or 10 to the power of 18.
1:24:06.1 DW: It’s robustly true that there are at least hundreds of millions of them. It’s robustly true that there aren’t 10 to the power of 100 of them, but there’s no robust answer in the middle. Now, I should confess, various of my more purist, metaphysically-inclined colleagues are tearing their hair at this point, because the metaphysics here is chaos.
[chuckle]
1:24:26.7 DW: But I think our metaphysics has to be responsive to our science, so to some extent, insofar as we need to change the way our metaphysics work to make sense of that kind of way of talking, that has to be changed in that way to do justice to science.
1:24:41.6 SC: And you seemed to have an implicit answer to this question in something you said earlier, but there’s the question of when the branching happens, how quickly does it spread, should we think of the branching as being simultaneous in some arbitrarily chosen reference frame throughout the universe, like when Alice and Bob are far apart with their entangled spins, and they’re going to measure them, when Alice measures hers, does Bob split instantly or is there some speed of light delay there, or is that a human construction, and we can choose it either way?
1:25:10.5 DW: Yeah, I mean, it’s kind of somewhere in between, and it’s a sort of trades on an ambiguity about what we mean by branches here, that doesn’t matter in small systems, but matters when you’ve got the Alice-Bob situation. So here’s something that doesn’t depend on how you set it up. The quantum state of the region of spacetime containing Alice will not change until light has had a chance to cross the gap from the region of spacetime describing Bob. That’s the fundamental, or more fundamental lower level description before we start talking about branches. I mean, there’s a bit of branching if we talk about a classical spacetime with Alice and Bob, but at the level we’re thinking about, we’re not thinking of branching it.
1:25:52.1 DW: If you describe it in branching terms, the way I prefer to talk would be to say that I want to think about a branch, my piece of emergent infrastructure as something that’s expanding on the forward light cone. So Alice branches her future, the branching spreads out from Alice at the speed of the fastest interaction that’s doing the branching. In practice, that’s always going to be light for big systems. Bob also has branching happening and eventually those light cones cross over and the Alice light cones… Sorry, I’m using my hands here, it’s a podcast, I’ll stop. The Alice light cone crosses Bob, the Bob light cone crosses Alice, and then Alice and Bob split again, so Alice and Bob experience two sequential splitting events, one when they did their own measurement, and one when they come into the causal future of the other measurement.
1:26:44.0 DW: But sometimes it’s convenient to talk more globally, if you want to write down a quantum state and say, we’ll break this state into Alice Bob both gets spin up, plus Alice Bob both get spin down, plus Alice gets spin up, Bob gets spin down, plus Bob gets spin up, Alice gets spin down. Then we’ll want to use a global notion of branching, but at that point, we are no longer picking out a sort of robust high-level feature any more, we’ve added a piece of pure convention, and the piece of convention is picked up by exactly by what you were saying earlier, that we’ve chosen an arbitrary reference frame.
1:27:18.7 DW: So we can talk that way now if you want to, but nothing… Nothing physically is happening instantaneously, even at the emergent higher level sense. There is something happening at the emergent higher level sense, but only on the forward light cones of the measurement. At least, this is the way I think about it.
1:27:37.1 SC: Yeah, no, I think it’s a perfectly… I think it is interesting because it is a kind of pressure point of emergence and subjectivity and objectivity and all these things. So again, for the youngsters in the audience, there’s work to be done in thinking about these things, it’s not all figured out yet. And we haven’t… To bring it all back and close the circle, we started this by saying that the quantum measurement process is a manifestly time-irreversible process. So how do you explain that, Mr. Everett?
1:28:14.0 DW: Right. Well, I don’t necessarily know the answer. I mean…
1:28:18.4 SC: What? You’re being given good money to be on this podcast.
1:28:18.5 DW: I’ll give you a classic philosopher’s way of solving the problem, which is by reducing it to a bigger unsolved problem. So claim, this branching is a special case of thermodynamic irreversibility. That goes back partly to what I was saying earlier about the the self-same equations happening in different contexts, but we can break it down more physically. Here’s the kind of argument you might give as to why Schrödinger’s cat causes everything to branch out from it. A random photon is coming in towards Schrödinger’s cat. It’s coming in from outer space, let’s say. If the cat is alive, the photon, then the cat’s standing up, the photon bounces off the cat. If the cat’s dead, the photon passes harmlessly past the cat and bounces off something different.
1:29:03.9 DW: And so now, the quantum state before the photon arrived was, let’s just say this is the very first photon is, cat alive plus cat dead, photon coming in, the quantum state afterwards is cat alive photon bounced off cat, plus cat dead photon missed cat. And this is an entangled state, it’s a joint state of both of them, the photon is now recording the quantum state of the cat, and then the idea is millions more photons come in and to a higher and higher degree, the cat just thoroughly entangled all these photons. Okay, but suppose we ran that process backwards. The backwards description of what’s going on would be something like, initially, the photon is entangled, quantum mechanically tangled up with the cat, and so the total quantum state is something like photon coming on a one trajectory dead cat, plus photon coming on a different trajectory live cat.
1:30:00.8 DW: And if you evolve that through, the photon coming in on the live cat trajectory bounces off the live cat, the photon coming on in the dead cat trajectory keeps going. And now I’ve got a single photon coming out, no longer entangled with the cat. And if every single photon was coming in in quantum states like that, then at the end of all that process, the cat would be in its… The cat would have started in this ridiculously entangled state; at the end of the process, the cat would have been completely pure.
1:30:29.6 DW: So why do we think one of those happens rather than the other happens? Well, in some sense, we think that that incredibly delicately entangled state of cat among the million photons is just a ridiculously implausible, ridiculously unlikely way for the system to have started off, whereas the state was just all the photons coming in any which way, that was a perfectly natural reasonable state for the systems to have started in, but as I’ve kind of shown, that’s an explicitly time-directed assumption. The state that I have said is completely outrageous, it’s just the state, it’s just the reverse of the state the system will in fact get in to in the future. The state I said was perfectly normal was one that I would have got into if I ran the ridiculously outrageous state backwards.
1:31:10.2 SC: And all of this is exactly analogous to the likelihood of the atoms in a glass of water being ready to make an ice cube form spontaneously.
1:31:21.4 DW: Absolutely. I’d say it’s not even analogous, it’s the same thing, it’s just being applied to a different problem. There are some slight quantum subtleties, but basically it’s the same thing. So the very same things that drive irreversibility in the ice cube drive the fact that systems branch rather than re-merging. I mean, going back to a thing you said right at the beginning is, do I different arrows of time, this is one of the reasons I don’t, really. If you start talking about the quantum arrow of time and the statistical mechanical arrow of time, you kid yourself into thinking these are different things.
1:31:49.8 SC: Well, to non-Everettians, they might be different things, right?
1:31:52.7 DW: That’s fair, that’s fair, although, it also shows why Everett is…
1:31:57.6 SC: Better. [laughter]
1:31:58.0 DW: Better, yeah, actually. It shows why there are awkwardnesses in not being Everettian, let me put it that way, because these things about the way systems naturally evolve and decay are happening anyway. It’s kind of weird if separately from that you’ve got an entirely different mechanism that really imposes the direction. If you have some different explanation of the arrow here, you have to ask, how does that mesh with the decoherence story.
1:32:24.9 SC: Because it’s going to happen anyway, yeah.
1:32:26.6 DW: It’s empirically tested. I mean, not the cats, but for mesoscopic systems.
1:32:30.5 SC: So just to restate that, the thing about Everett, whether it’s the fact that there is branching or it fits into this bigger statistical mechanical picture, is that all of the processes you need to make Everett work will also happen in everyone else’s version of quantum mechanics or statistical mechanics or whatever, but then they’re saying, and I have extra processes also ’cause it makes me feel better somehow about the explanation.
1:32:55.2 DW: Yeah, I can make the case for a defense against that, but yeah.
1:33:00.7 SC: I know. We’re among friends here. But okay, so, and I’ll get back to a difficult question then, how do you think about the past hypothesis in these quantum mechanical terms, is it just one past hypothesis for the early universe and its wave function, or do we have to be more subtle about different senses in which the early universe had low entropy?
1:33:23.9 DW: Well, this goes back to why I would rather not place the past hypothesis as a statement about low entropy. I would rather put the statement about not having lots of delicate correlation entanglement. I think the past hypothesis is a kind of cosmic version of the hypothesis that all the incoming photons were not complicatedly entangled. If our past hypothesis that the initial quantum stage was a relatively simple state without a bunch of complicated entanglements, that’s probably mostly what we need.
1:33:54.1 SC: Good, so actually, that’s… In fact, that’s perfect because we can wrap up the conversation by going back to the weeds that we were in a little bit ago that I said let’s defer that. I think we can do it now because, let me try to say, if I understand what you’re saying in slightly different words, then you can tell me whether I’ve gotten it completely wrong or not, there’s a variant of the past hypothesis that just says, the early universe had low entropy… Oh, and by the way, it was in a low entropy state that within its macro state, it was pretty generic, it wasn’t just low entropy, it was low entropy without any secret conspiracies that was going to make it even lower entropy, right. That’s one version of the past hypothesis.
1:34:33.0 SC: And there’s another version that says, look, we look and we see the cosmic microwave background in our telescopes. It’s true that if the question you asked was, of all the ways you could have seen that, what do most of them look like? And what… Of all the ways the early universe could have been to give us this image in our telescopes, what do most of them look like? And most of them look wildly inhomogeneous with crazy fluctuations and gravitational differences from here to there, but there was structure in the specific micro state that canceled all that out so that we see smooth radiation in our telescopes. And your point, if I understand it is, sure, if the question you’re asking is just what do most such early universe states look like conditioned on what we see, that’s the answer you get.
1:35:24.4 SC: But if you add to that a requirement that what do most early universe states that didn’t have crazy correlations in them and conspiracies look like, that would give us our state today, then you automatically can derive the fact that, oh, it had to be pretty low entropy in our coarse grain sense. Was I close there?
1:35:44.6 DW: Yeah, that sounds right, yeah. Put it this way, if the early universe was in a non-crazy microscopic state, then something like the second law holds and the macroscopic dynamics for that universe have entropy going up. And once you’ve got the macroscopic dynamics have entropy going up, then obviously the entropy of the early universe has to be lower than the entropy of the present universe.
1:36:04.9 SC: Good. Yeah, so I think that I have not really sat and chewed on that different formulation much myself, and I’m not objecting to it, I’m not philosophically opposed to it, I think it makes perfect sense. And in fact, it leads to… I already said this is the final thing, but we have another final thing now, it leads into this… The questions that arise in the other direction of time, we’ve been talking about the past, but we also have the future, there are some of my best friends who worry about the fact that in the future, we have a positive cosmological constant that is making the universe accelerate, everything will eventually empty out, and we’re in what looks like equilibrium in some sense, just empty space. De Sitter space is the technical term, but de Sitter space has a non-zero temperature according to Stephen Hawking and Gary Gibbons in the 1970s, so there could be random fluctuations.
1:36:57.0 SC: And if you wait a very, very long time, you will get fluctuations into Boltzmann brains or even entire Boltzmann galaxies or universes and so forth. So are we worried about getting back to the recurrence objection that Zermelo gave to Boltzmann years ago where if you say the universe started 14 billion years ago, but lasts infinity years to the future, who cares about the first 14 billion years, most of the occurrences of people like us will be random fluctuations in the future. Is that something that you personally lose sleep over?
1:37:32.9 DW: Lose sleep overstates it, but I think it’s a serious concern. I think it’s a little subtle whether it’s true, partly for reasons about what fluctuation really means in an Everettian universe that you’ve talked about, partly because ultimately, this is probably a physics that is tentative enough that we should be cautious about it. But yeah, notwithstanding all of that, there is something concerning about it, I think, if the universe is such that it predicts that it’s full of beings who have not even fake histories, but entirely true histories that come from fluctuations of their own galaxy to existence or something. I think there’s some reason to be worried that the theory’s self-undermining.
1:38:26.5 DW: In a certain sense, there’s a bunch of problems I think that come up in physics where… We use infinity pretty casually in physics and mostly we use infinity as a convenient approximation for large finite things. I mean, particle physicists talk about collisions between particles and then we see what happens at infinity, and some philosophers, as you know, better worry about exactly how you formulate those infinities, but as my colleague, Porter Williams, says, the main tube in the LHC is like about three meters wide and whatever “at infinity” is, it’s much less than three meters. So in a case like that, all we really mean is something like much bigger than the scales we’re interested in.
1:39:10.3 DW: It’s easier to approximate things by infinity than by some large finite number. Every now and then a real infinity turns up, and when our theory has a real infinity, in this case, the literally infinite future of the de Sitter universe, then I sort of think at some level, I’m not certain we know how to theorize about things of that kind. And that’s not to say, therefore we should be relaxed about the Boltzmann brains that you talk about, it’s more the opposite to say, this is a reminder to us that in positing something that’s really infinite, whether it’s infinite into the timelike future or infinite into space, it’s not that that posit is illegitimate, but that posit is a very substantial, deep thing that goes beyond our mathematical use of infinity in other places, and we’re right to kind of pause and worry about it.
1:40:00.9 SC: Yeah, I think that’s actually very much in line with my own view on these things, it’s not a show-stopper for anything, it’s a reminder there are things we don’t understand, and we shouldn’t be too glib about them. Some of my colleagues in cosmology think that it’s a reminder that you’ve lost your mind that you worry about these things, so for the truly very final answer, are you mostly encouraged or mostly discouraged by the relationship between philosophy and physics these days? Do you think that there’s been more overlap that’s been constructive or do you spend more of your mental energy sort of frustrated that it’s not even better?
1:40:38.0 DW: I’m mostly positive about it. I sometimes think that question reflects more people’s natural disposition than their objective judgment of the field. My feeling is the philosophy of physics has upped its game and has, or bits of it at least, have increasingly engaged with questions in physics that genuinely bother physicists. And I think conversely, it’s not true that physicists are not bothered about conceptual problems, they’re not bothered about conceptual problems that are seen as sterile, and I think the quantum measurement problem has traditionally often seemed sterile to physicists. It doesn’t seem to lead to something, but without having time to talk about what this is, I imagine your readers, your listeners have come across, I mean, the black hole information loss paradox has had the finest minds in theoretical physics tearing their hair out over it for the better part of 50 years, and it’s a conceptual problem, it’s not a calculation issue, so it’s not that these things don’t matter.
1:41:48.3 DW: And I think if you can come to the table in that conversation with something to contribute, then I think you can relatively easily get across the prejudice that can occur. I think you do run into genuine prejudices. I think it helps to say the word conceptual rather than philosophical. I think it helps to have a physics PhD, more because you could say I have a physics PhD than because of any concrete knowledge you got from it. But I think it could be done. I think there are good conversations to have, and a lot of people are willing to talk. And I think, because I’ve mostly put that in terms of what are the issues of engaging with physicists, I think it is true that the philosophy of physics has some bad habits of its own here, that it can help get over.
1:42:32.4 DW: I think the philosophy of physics has this slightly awkward situation where mainstream physicists are not naturally inclined to talk to them, but people in much more niche areas of physics who are struggling themselves to get physicists to talk to them, are much more willing to talk to philosophers, and so philosophy of physics to a degree that concerns me sometimes can often be a philosophy of dissident physics, or minority physics, which isn’t to diss any of that, I mean, we don’t do science by majority vote, but it does often mean that the center of gravity and energy in the philosophy of physics often struggles to be on the conceptual questions that physicists really care about.
1:43:16.1 DW: So yeah, that’s a kind of cautious optimism. I think there’s still problems, I think there are institutional structural problems, and I also think, bluntly, many people doing philosophy of physics, or who want to do philosophy of physics, aren’t necessarily always doing the homework to actually get on top of things to the degree that they don’t make relatively elementary technical errors that will be seen correctly as invalidating what they’re doing by physicists they’re engaging with. So ultimately doing interdisciplinary work is hard, you have to collaborate, you have to talk other people, you have to put the time in, and there’s always a prejudice on both sides against it that you have to deal with. But yeah, I’m kind of optimistic. I think the best days are ahead of it, for that point of view.
1:43:57.2 SC: You did a good job of being optimistic, but then being very cautiously optimistic, but we will take that for what we’re looking for here. So David Wells, thanks very much for being on the Mindscape Podcast.
1:44:06.2 DW: Thanks for having me.
[music][/accordion-item][/accordion]
Great Podcast Sean!
Two points:
Past leaves chain of traces that are available at present moment, future does not. If foot imprints on the beach get washed away, we will not know someone walked on the beach in the past even though entropy increased. So is it as simple as past is different from the future not because entropy increases but because past events tend to leave traces that can be accessed (remembered) in the present?
It is generally said that microscopic physics is time reversible. However is it more correct to say that the microscopic physics is velocity reversible? Velocity is a vector so it has direction and opposite velocity is reverse of for forward velocity. And that opposite direction provides the negative sign not the time. I say this because even in the classic, intuitive example of time reversibility of microphysics that is given i.e. running the film of billiard balls bouncing off of each other backwards, watching it one is not able to tell the difference that film is running backwards or forwards. Well, we are running the film in reverse direction still in forward time. It is just that we ran the film playing mechanism backwards which for mechanical projector could be the motor running in reverse direction or digital projector sampling the digital data in reverse order relative to the order it was written into memory?
In the simple example of billiard balls, we have single mechanism to run the film in reverse. However to practically make that happen a daemon will have to reverse the velocity of all particles everywhere at the same time (simultaneously?) which I guess a daemon can do, even though simulteneity is not supported by GR.
There is a macroscopic physical process that seems to be truly irreversible ( not only ” in practice” ) , and that is Gravitational Collapse :
If a 10 solar mass ball of iron dust collapses to a black hole and , subsequently , evaporates via Hawking radiation , dissipates all its energy in the surrounding environment , and leaves behind nothing else.
The problem , now , is that the reversed procedure is not physically conceivable at all:
One has to start with some region of spacetime that absorbs thermal radiation and creates a tiny white hole ( but , why not a a tiny black hole? It would have been the obvious result of such a concentration of radiation from thermal fluctuations!).
Then the tiny wh has to grow to a 10 solar mass macroscopic one ( but , how? The white hole antihorizon does not consume anything from outside , being a ” past horizon” , so ,the only way it has to grow is to Hawking-radiate negative energy photons or other particles , and that does not make much physical sense!).
This seems to be the only way that a white hole could gain positive mass ( from the ingoing positive energy partners of the Hawking rad.), and that just seems impossible!
And I leave aside the other issues such as: the instability of the wh antihorizon , or that the “final object” has to be an iron ball of dust and not something else.
Am I missing something important?
Another wonderful discussion! Thank you both. I may be missing some big or fundamental things here, being a biologist. Doesn’t the whole multiple worlds thing hinge on the idea that the wave function collapses in our branch vs a hidden variable where it was predetermined, we just didn’t know the answer? And the assertion that Bell’s inequality has been violated, proving no hidden variable? But if we successfully challenge this assertion – doesn’t the whole argument fall apart? Very few folks discuss the actual experimental limitations of the Bell test experiments, and like a magician, redirect attention to an “interesting” issue rather than the core issue, resulting in most physicists claiming the Bell test result is beyond reproach. If we assume there has yet to be a conclusive test, then does this all become moot?
Related to my previous comment:
I’m aware that some of the issues with the time-reversed version of Gravitational Collapse-Evaporation process that I mentioned , have to do with fine-tuned initial conditions (e.g. how we avoid the white hole Novikov instability or why the wh has to grow until it reaches 10 solar masses? Why not 10 billion?
Why has to stop growing at all?).
Some other issues are probably related with the information loss problem ( why iron dust?).
But the first issue that I mentioned in my previous post has to do with a , seemingly , impossible physical mechanism , not with the 2nd law , so what’s going on?.
If the universe wasn’t expanding would the second law of Thermodynamics still apply?
What would happen with the 2nd law If gravity took over in the universe and it starts Contracting? Is the second law ultimately Defined by the mere existence of dark energy or it can be explained by other fundamental arguments?
Ok , I think I figured it out ( partially):
It seems that all we need is to interpret this semi-classical process in a way that makes physical sense.
The “negative energy” Hawking radiation from the white hole is physically equivalent with the absorption of positive energy thermal radiation from past null infinity , as the wh antihorizon grows.
A future observer cannot directly observe this , so , instead , she thinks that the white hole emits , effectively , negative energy radiation towards the future null infinity. This is not a problem as the total mass/energy is positive:
Initially , before the wh forms , we have only thermal radiation , with 10 solar mass equivalent energy ( in my example). As the w hole grows , the radiation is consumed and , finally , it disappears when the wh reaches its maximal mass. Then the wh “explodes” and releases the iron dust , assuming that , quantum mechanically , the process is reversible.
There are still some things that seem problematic , and I’m not entirely sure that I got it right.
Anyway , the podcast was great and thought provoking ! Thanks!
Pingback: Sean Carroll's Mindscape Podcast: David Wallace on The Arrow of Time | 3 Quarks Daily
I’m no expert, but if I had to choose my favorite hypothetical theory that attempts to explain the arrow of time, it would probably be a multiverse based on de Sitter space and an idea suggested by Edward Farhi, Alan Guth, and Jemal Guven called ‘baby universes’. Sean Carroll ends his book ‘FROM ETERNITY TO HERE: THE QUEST FOR THE ULTIMATE THEORY OF TIME’ with the following paragraph:
‘The nice thing about a multiverse based on de Sitter space and baby universes is that it avoids all the standard pitfalls that beset many approaches to the arrow of time: It treats the past and future on equal footing, doesn’t invoke irreversibility at the level of fundamental dynamics, and never assumes an ad hoc low-entropy state for the universe at any moment in time. It serves as a demonstration that such an explanation is at least conceivable, even if we aren’t yet able to judge whether this one is sensible, much less part of the ultimately correct answer. There’s every reason to be optimistic that we will eventually settle on an understanding of how the arrow of time arises naturally and dynamically from the laws of physics themselves.’
While there is no way to know for certain if universes other than our own ‘observable universe’ exist, it seems unlikely (at least to me) that the kind of processes that were responsible for the universe we inhabit to come into existence would only have occurred once, never to be repeated.
Maybe if Schrodinger had taken his own deterministic equation more literally, like Everette did, he could have applied his vast genius to demystifying much of the mysterious behavior that plagues quantum mechanics.
@HowardGary Actually Schrödinger did take his own deterministic equations seriously, and Heisenberg in his book Physics and Philosophy did respond to his attempt. And Everett explicitly responded to the points raised by Heisenberg in his thesis. This proves that he was well aware of this predecessor of his own work, and of the objections that had been raised against it.
Compared to how unfriendly Heisenberg reacted to Einstein’s objections and Bohm’s proposal in that same chapter, his assessment was actually quite sympathetic (probably because he appreciated Schrödinger as a person): “Among the remaining opponents of what is sometimes called the “orthodox” interpretation of quantum theory, Schrödinger has taken an exceptional position inasmuch as he would ascribe the “objective reality” not to the particles but to the waves and is not prepared to interpret the waves as “probability waves only.” In his paper entitled “Are There Quantum Jumps?” he attempts to deny the existence of quantum jumps altogether (one may question the suitability of the term “quantum jump” at this place and could replace it by the less provocative term “discontinuity”). Now, Schrödinger’s work first of all contains some misunderstanding of the usual interpretation. He overlooks the fact that only the waves in configuration space (or the “transformation matrices” ) are probability waves in the usual interpretation, while the three-dimensional matter waves or radiation waves are not. The latter have just as much and just as little “reality” as the particles; they have no direct connection with probability waves but have a continuous density of energy and momentum, like an electromagnetic field in Maxwell’s theory. Schrödinger therefore rightly emphasizes that at this point the processes can be conceived of as being more continuous than they usually are. But this interpretation cannot remove the element of discontinuity that is found everywhere in atomic physics; any scintillation screen or Geiger counter demonstrates this element at once. In the usual interpretation of quantum theory it is contained in the transition from the possible to the actual. Schrödinger himself makes no counterproposal as to how he intends to introduce the element of discontinuity, everywhere observable, in a different manner from the usual interpretation.”
@GENTZEN I stand corrected it does seem Schrodinger did take his own deterministic equations seriously.
When Schrodinger first presented his cat paradox in 1935 it is clear he had already considered the possibility that there is a dead version and an alive version of the cat. He just thought it was absurd at that time, as a problem of quantum mechanics that we have to solve, rather than the solution.
Later on is seems he had somewhat of a change of mind. In Dublin in 1952 Schrodinger gave a lecture in which at one point he jocularly warned his audience that what he was about to say might ‘seem lunatic’. It was that, when his equation seems to be describing several different histories, they are ‘ not alternatives but all really happen simultaneously’. This is the earliest known reference to the multiverse.
So while Schrodinger had the idea, and eventually gave it more credit, Everett rediscovered it and made it into a full-fledge theory. Much has been written, but little has been added to the many-worlds interpretation after Everette.
Ref: QUORRA ‘What evidence is there that Erwin Schrodinger was a proponent of the many-worlds interpretation of quantum mechanics?’
I would like to accept the many-worlds interpretation of quantum mechanics (MWI), mainly because it seems to answer so many otherwise unexplainable properties of our so-called ‘observable universe’. The one question that plagues me the most, and I don’t believe even MWI has any hope of resolving is: Is the particular universe we inhabit, our observable universe, the result of a measurement (interaction) in some other universe, or is that ‘other universe’ the result of some measurement that took place in our observable universe?
In a way it reminds me of past debates that went from the Earth being the center of the universe, to the Sun being the center, to our galaxy being the center, to there being NO center to the universe. Now, if one accepts MWI at face value (i.e. all these universe actually exist) the question seems to be: ‘Is there one particular universe that started it all off?’ – The ‘Master Universe’. That may seem like a ridiculous conjecture, but so is the notion that every universe is the cause of every other universe.
Can anyone help me on this one?
@HowardGary Interesting, so Schrödinger got even closer to Everett’s ideas than his paper “Are There Quantum Jumps?” suggests.
After recently reading Bohr’s response to EPR, I wondered whether he ever worked out the “obvious” quantum physical consequences of his principle of complementarity. The complementarity between momentum and position have been discussed to death, in the form of wave particle duality, double slit experiments, particle tracks in cloud chambers, and all that. But what about the complementarity principle for time and energy? Preparing a system in a thermal state is not difficult, after all quantum physics began by an analysis of black-body radiation. Since measurement and preparation belong together, what would have been the corresponding measurement? The measurement of the spectral lines of atoms and molecules? Or measurement of any other non-randomly fluctuating macroscopic observable?
The thermal state of the early universe also occurred in the discussion between David and Sean, and David argued against making a big deal of mystery about it: “So it’d really better be the case that the early universe was lower entropy than the present universe. We shouldn’t need to give some subtle transcendental argument as to why it must be lower. It’d better be lower in order not to blow our entire thermodynamic theory out of the water.” My impression was that David really has a deep understanding of those conceptual physical topics.
@GENTZEN Sometimes overshadowed by more well known equations like Newton’s F = ma, Einstein’s E= mc^2, Planck’s E = hf, etc., the second law of thermodynamics which states ‘ When a system containing a large number of particles is left to itself, it assumes a state with maximum entropy, that is, it becomes as disordered as possible.’ This simple sounding statement may top all others in importance in the role it plays in the formation and evolution of the Universe. According to renowned astronomer Arthur Eddington, “The law that entropy always increases, holds, I think, the supreme position among the laws of nature.”
MWI, if true, could help resolve some of the deepest mysteries of the cosmos, at both the macroscopic and microscopic levels. The ‘good’ thing about MWI is that it can’t be disproven, the ‘bad’ thing about MWI is that it can’t be disproven’. So is MWI a scientific theory (actually it’s an interpretation of quantum mechanics) or not ? It all comes down to how one chooses to define a ‘scientific theory’.
@HowardGary Whether or not MWI can be disproven depends on how you present it and where you put your focus. The way David Wallace has introduced it here seems careful enough such that it could only be disproven by actual physical experiments observing wave function collapse (or maybe by lessons quantum gravity): “The right way to go, I think, is something like the many worlds interpretation. We want to say that quantum mechanics really maintains all those superpositions, and the fact that it looks as if we’ve got one of the possible outcomes at all there at once, rather than all of them at the same time, is perspectival. It’s that we just happen to be the kind of physical system that’s sitting in one of those terms”
The observation that we happen to be the kind of physical system that’s sitting in one of those terms seems to remain true, independent of whether other versions of us happen to sit in others of those terms or not.
As long as we can observe interference effects from those other terms, we might even claim in a certain sense that we can prove their existence. But on the level of gravity, we no longer can. There are those like Freeman Dyson who explain why quantizing gravity might be a bad idea. But Sean Carroll too does not want to quantize gravity, but instead he wants to get it as an emergent phenomenon from MWI itself. So even lessons from quantum gravity are only good enough to disprove certain presentations of MWI and unwarranted conclusions, but not MWI itself.
Now popular level presentations of MWI are a different beast. In a certain sense, Sabine Hossenfelder made a video disproving MWI shortly after Sean’s book on MWI appeared, and she certainly had read that book. But the version of MWI disproved by her has nothing to do with the version introduced by David Wallace. The question of whether Sean’s book sufficiently distanced itself from the version disproved by Sabine Hossenfelder is not really fruitful. I find it more fruitful to look at attempts to disprove certain unwarranted conclusions from MWI.
Um dos episódios mais interessantes.
As diferenças inteletuais sobre, “inflamaram” o entusiasmo. Ótimo, quando confronto de ideias.
Sugiro, se possivel, e, quando oportuno, um outro episódio.
Obrigada.
@GENTZE It may be that one way to disprove MWI is by actual physical experiments observing wave function collapse. But how would one, in in theory, go about devising an experiment to prove wave function collapse? My personal belief is that MWI is so elegant and helps provide so many solutions to so many problems in cosmology and particle physics that it deserves the recognition it receives by many top physicists, even though (in my opinion) it’s extremely unlikely any experiment will ever be devised to disprove it.
I may have been a bit hasty in saying there is no way to falsify MWI. Supposedly there are many experiments that could potentially falsify quantum mechanics (QM) as a whole, and since MWI is an interpretation of QM that would also falsify MWI. But as far as I know there is still no way to falsify MWI without first falsifying QM.
@HowardGary Many experiments investigating wave function collapse have been performed over the years. A recent one from 2019 is described in An ultra-narrow line width levitated nano-oscillator for testing dissipative wavefunction collapse. This and other experiments are mentioned in section 7. CSL and Experiments of Giancarlo Ghirardi’s and Angelo Bassi’s SEP article on Collapse Theories.
Decide for yourself whether observing wave function collapse would falsify quantum mechanics (QM) as a whole. Since it is not obvious how to unify QM and GR, my guess is that experimental results probing their interaction would typically not be seen as falsifying QM or GR. At least not without further developments on the theoretical side.
@GENTZEN Thanks for the info about experiments investigating wave function collapse. As far as whether or not actually observing wave function collapse would falsify QM as a whole I agree with you that it wouldn’t. For example in the Copenhagen interpretation of QM, wavefunction collapse occurs every time a measurement (interaction) takes place.
The remaining question is whether or not evidence of wave function collapse would invalidate MWI? On the face of it, it would seem to, since MWI asserts that the wave function never actually collapses into one definite state when a measurement takes place, but all possible outcomes occur, each possible outcome taking place in one of a multitude of other universes. But in defense of MWI, evidence of wave function collapse is NOT proof of wavefunction collapse. I could be wrong, but I’m pretty sure for any claim that wave function collapse has taken place an alternative explanation could be found.
@Howard Gary The main difficulty about these experiments is to distinguish between actual objective collapse and good old decoherence, but people try to find ways to overcome this.
These experiments are important, because any observed deviation from the Schrödinger linear equation, will potentially have many implications not only for the foundations of QM (the measurement problem etc), but also for Quantum Gravity, the black hole information loss problem and other related conundrums.
@DIMITRIS I have no objection to experiments investigating wave function collapse. As you say “Any observed deviation from the Schrodinger linear equation, will potentially have many implications not only for the foundations of QM (the measurement problem etc), but also for Quantum Gravity, the black hole information loss problem and other related conundrums.” You also state “The main difficulty about these experiments is to distinguish between actual objective collapse and good old decoherence.” That may be one of the most difficult experiments ever, but if they can successfully achieve this it would no doubt be one of the greatest, if not the greatest, accomplishments in the history of experimental physics.
The video ‘How Decoherence Splits the Quantum Multiverse – YouTube’ does a pretty good job of explaining why it is so difficult to experimentally distinguish between actual wave function collapse and decoherence.
Why is the past different from the future? Sean Carroll explains how the arrow of time is not an intrinsic property of physics, but rather an emergent feature in the YouTube video:
‘The Arrow of Time feat. Sean Carroll’ (13 Nov 2011)
Both Sean Carroll and David Wallace clearly express important issues concerning the problem of understanding the direction of the flow of time. Both agree that there is some relationship to entropy or at least that entropy plays an important role in understanding the flow of time. As naïve observers, the correlation between time going forward (direction but not magnitude), and changes that appear irreversible, is obvious in our world view. Yet there are problems with such a viewpoint. In particular, there is the question about whether entropy is the right lens to examine time. For me there are three questions that need to be better understood.
1. From standard astrophysics’ analysis, the early universe started as homogeneous and isotropic and expanded from a Planck scale following a constant entropy trajectory over tens of orders of magnitude in time. How then can we say that entropy increase is critical to the forward flow of time?
2. If entropy is a cause of the flow of time, then, which entropy increase are we actually talking about? Entropy locally can increase, decrease, or whatever and the flow of time does not change in that locality. Surely that doesn’t work. How about the increase in average entropy of the universe? Here the direction of the flow of time works poorly quantitatively (one of the key requirements of a good theory). It also brings back “action at a distance” (and only after averaging over time and space). That is not very satisfying.
3. Almost all of the fundamental equations of physics are time symmetric invariant (replacing t with -t does not change those equations.) We only have to look at how we plot things to know that +t goes to the future and -t goes to the past. Quantum Mechanics seems to suggest otherwise. Each time we start from some initial condition, whether to the future or past we observe that the actual path cannot be predicted on a fine scale.
What is the ultimate fate of the Universe? It depends on the so called ‘critical density’ (the average density of matter required for the Universe to just halt its expansion, but only after an infinite time). A universe with the critical density is said to be ‘flat’.
There is a growing consensus among cosmologists that the total density of matter is equal to the critical density, so that the Universe is spatially flat. Approximately 29% of this is in the form of low pressure matter (5% ordinary baryonic matter plus 24% non-baryonic “dark matter”), while the remaining 71% is thought to be in the form of a negative pressure “dark energy”, like the ‘cosmological constant’ (the energy density of space, or vacuum energy, that arises in Albert Einstein’s field equations of general relativity). If this is true, then dark energy is the driving force behind the fate of the Universe and it will expand forever exponentially.
Ref: WMAP – Fate of the Universe
The second law of thermodynamics may be formulated by the observation that the entropy of isolated systems left to spontaneous evolution cannot decrease, as they always arrive at a state of thermodynamic equilibrium, where the entropy is highest. An increase in entropy accounts for the irreversibility of natural processes, often referred to in the concept of the arrow of time.
Ref: Wikipedia ‘Second law of thermodynamics’
In practical situations the main problem is in determining if the system you are interested in is an isolated system or not. It is extremely difficult, if not impossible, to insure that any system is entirely isolated from its surrounding environment, much less trying to make a measurement of that system, without disturbing its isolation. So how would one go about verifying that the entropy of isolated systems left to spontaneous evolution cannot decrease? I guess that’s why the second law of thermodynamics is called a ‘law’ not a ‘theory’. It’s one of those postulates that we simply have to accept (or not accept) even though in all likelihood it can never be proven (nor disproven).
Suppose, just for the sake of argument, that the entropy of an isolated system could decrease, does that imply the arrow of time could flow in both directions? And if it could does that mean the future is determined in the same way we assume the past is?
The relationship between the direction of time and entropy, where entropy is given as a reason or explanation for the arrow of time may be wrong. Such reasoning is neither quantitatively nor, I believe, even qualitatively true based on what we know about the universe. Still, there are intriguing correlations that need to be accounted for. It may all reduce to the possibility that we have confused what is the cause and what is the effect. Perhaps it is the flow of time that makes the second law valid. Perhaps that is why the time symmetric Boltzmann equations work in both time directions. If so, the flow of time may best be understood as the law-like logical next steps. It is what counts as an explanation rather than the sign of “t”. Laws such as the law of inertia are built on logical (in the mathematical equation sense) next steps and is totally independent of the sign of t. Consistency and actually reflecting the world should be the ultimate criteria for explanation. Entropy as an explanation for the direction of time would seem to lack both.
In the video ‘Your Daily Equation #32: Entropy and the Arrow of Time’ (3 Jun 2020), theoretical physicist Brian Greene describes some of the reasons he believes the best explanation for the so called ‘arrow of time’, the fact that we always witness some events taking place in a certain order, but never in the reverse order (e.g. glasses shattering, but never un-shattering, candles burning, but never un-burning, etc.) is because of the so called ‘past hypothesis’ , where the very early universe was, for what ever reason, in a very highly ordered, low entropy state. And ever since there has been an overwhelming tendency for systems to evolve from ordered states to less ordered configurations, but not to go in the opposite direction. This tendency is spelled out in the Second law of thermodynamics.
Of course that leaves the question of why the very early universe was in such a well ordered, low entropy state to begin with, but once we accept the fact that it was, the rest seems to make perfect sense.
Back to the fundamentals. Try this Gedanken experiment.
We have 5 particles in a box and imagine there are 2 smaller sub-boxes. We then calculate the entropy of the configurations of that system. The amount of time the system goes (in the next step) towards a higher entropy is not that much greater than the amount of time the system goes towards a lower entropy. If you are watching and calculating you will see a lot of next steps going to a lower entropy. It is not negligible. Locally then there is a poor directional correlation between increasing entropy and the subsequent direction of time. This explains our psychology when we look at (what we think of as) statistical tendencies.
The correlation for the Boltzmann statistical entropy for many many particles is, of course, much much better for entropy increase and the direction of time. Would you want to say that only for large numbers of particles is the collision dynamics correlation causal but otherwise not? Hardly a strong theory. But it does make a good story. The fall back position for many who hold this entropy view is that it is only the direction that counts and then only as a correlation with the time average of the universe and only then when you get the right answer. Why not accept that the causal story goes the other way. It is the direction of time which provides reason for entropy to increase.
Emergence and other approaches to explain the direction of time also fail if it is the case that causality goes in the other direction. My take on this problem contradicts the standard “Just So” story. So it goes.
If you had to choose one thing in the physical world that MUST EXIST in order for everything else to exist, would it be ‘space’, ‘energy’, ‘matter’, ‘time’, or some other entity? I think it would be TIME , because given enough time anything that could possibly exist, will eventually come into existence, but that’s just my opinion.
I’m frustrated with David Wallace because he seems to be arguing that the reversibility of most physics equations somehow means that there is no explanation in thermodynamics for a direction of time. That’s true but nobody is arguing that point. He’s creating an argument with a proposition that he has created for himself.
The direction of time doesn’t come from thermodynamics and isn’t itself an interesting or remarkable thing. Time is just one dimension in space-time and you know which direction you are moving in it by reference to whether you moving towards or away from the Big Bang.
Of course in reality we don’t move in time at all. There is no arrow which moves and all things just exist in four dimensions. As humans it feels like are moving in time but that is an illusion due to the fact that we cannot remember the future. That in turn can be explained by thermodynamics.
The fact that Newton’s laws of motion are reversible doesn’t invalidate Boltzmann’s explanation of the second law at all. If you take a low entropy arrangement of particles with random position and momentum then there is a high probability that when you move in the direction of time away from the Big Bang (I.e. forwards) that entropy will increase. If you stop the clock at that point then yes the equations are reversible and you can go back in time to your low entropy state. BUT if you take a random high energy arrangement of particles are run the equations backwards going back in time then it’s almost impossible to move to a low entropy state.
No contradiction no problem. Stop wasting everyone’s time David.
The conversation with David Wallace touched upon different interpretations of quantum mechanics (including the Many-Worlds (Everett) interpretation). An interesting topic for a future interview would be a discussion about superdeterminism.
A superdeterministic theory is a ‘hidden variable theory’ that solves the measurement problem and
a) reproduces quantum mechanics on the average,
b) is ‘deterministic’,
c) is ‘local’ in the sense of not having “spooky action at a distance’.
It’s a deterministic, local, collapse process that reproduces quantum mechanics exactly – no finetuning necessary.
It might even be able to be tested experimentally.
o For all superdeterministic theories identical measurement setups will lead to identical measurement outcomes.
o This is not the case for quantum mechanics!
o Look for autocorrelations in time-series of measurement outcomes that, according to quantum mechanics, should be uncorrelated.
o This generally requires small, cold, systems in which measurements can be repeated in rapid order.
Superdeterministic theories are not interpretations of quantum mechanics. They are more fundamental theories from which quantum mechanics derives.
Ref: Warsaw Spacetime Colloquim #11 – Sabine Hossenfelder (2021, 03, 26)