Transcript of Currents 016: Robin Hanson on Are We Living In A Simulation?

The following is a rough transcript which has not been revised by The Jim Rutt Show or by Robin Hanson. Please check with us before using any quotations from this transcript. Thank you.

Jim: Today’s guest is Robin Hanson. Robin is a regular on the show. I think this is his fourth appearance. In his day job, Robin is a professor of economics at George Mason University. He described his mission modestly as a secret quest to understand everything and to save the world. Pretty cool, I like that.

Robin: Looking at my progress and how long I’m liking to live, I’m not really sure that that’s really a target, but we’re going to keep trying.

Jim: We’re going to keep trying, all we can do right. He writes an amazingly interesting blog at overcomingbias.com and his tweet stream @robinhanson is well worth following. Today’s topic is Do We Live in a Simulation? A topic that I remember discussing in a freshman bullshit session at MIT in 1971. People have been thinking about this for a long time. Robin’s been thinking about it, in a formal sense, for longer than most. Before we jump into it, I’m going to stipulate, at least Robin can push back on this if he wants, I’m going to stipulate that there are limits to this kind of speculation about the ground of being that we’re going to get engaged in today.

Jim: For example, we can’t disprove the assertion that the universe which created five seconds ago with all of our memories in place and all ballistic objects in motion, so there are serious bounds to metaphysical speculations of all sorts, but with that, let’s hop in.

Robin: I got a degree in philosophy of science a long time ago, I think I’m … I did AI work which is a lot of how you do reasoning and thinking. After all that background, I’m relatively skeptical of the value of introducing these fundamental limits on what you could know arguments. It seems to me you just dive into a topic and you figure out what you can say and what you can’t, temporarily at least, and that’s what you should do. Trying to draw some grand law about topics you should not go into because you couldn’t possibly know just doesn’t seem to work for me.

Jim: I agree. I said, we’re going to live in the actual world. The fact that we can’t disprove logical impossibility of X, Flying Spaghetti Monsters or the universe was created five seconds ago. It can constrain us. Otherwise, we’d starve to death, right? That’s the reality of a pragmatic being in the universe, is we make decisions, we have ideas, etcetera. Back to the topic of Do We Live in a Simulation? The earliest thing I found from you in writing at least was a paper back in June of 2001, which I believe was prior to Bostrom’s speculations, very interesting and in some ways humorous paper called How to Live in a Simulation? Was that the start of your thinking or have you been thinking about this for longer?

Robin: I had been on mailing lists which had discussed the topic for the previous 10 years as had Nick Bostrom. He didn’t get that out of his own head. That was also an idea that lots of people have been talking about for a while, hence the mailing list that we shared. I was responding in part to people talking about it a lot, but I find it interesting that for odd hypotheses like this, there’s vastly more interest in the question, “Is it true? Then how does it matter?” Even like, “Are there aliens?” or other sort of dramatic hypotheses, overwhelmingly, people will talk about, “Is it true?” and they rarely get around to talking about, “How does it matter if it matters?”

Robin: My paper was there on the, “How does it matter if it matters?” topic and it seemed to me that that was neglected, so I was able in a relatively short time to come up with a bunch of obvious conclusions about how it would matter and that would affect what you would do if you believed you lived in the simulation. Similarly, you could write other analyses on what the consequences are if there are aliens or if we’ll be able to do brain emulations in the future or if there were long ago advanced civilizations that died down that we don’t yet know about, all these topics. I find it interesting that people have an overwhelming interest in talking about, “Is it true?” and you get to, “Okay, if it were true, how would it matter?” people lose interest.

Jim: It’s interesting and you came up with a quite snappy argument on what you should do if we do live in a simulation. In fact, I’m going to read the quote back from your paper, which I just thought was hilariously rich and good. I don’t know quite what to make of it, but I’d love to get your thoughts on it. “If you might be living in a simulation, then all else equal, you should care less about others, live more for today, make your world look more likely to become rich, expect to and try more to participate in pivotal events, be more entertaining and praiseworthy and keep the famous people around you happier and more interested in you.” I just love that. I’m not quite sure why, but it’s just very interesting.

Robin: Well, I would say it is the logical implications of the hypothesis. What might be funny is that you usually expect advice people give to fit the usual tropes of admirable living. Usually, whatever arguments you give are supposed to draw the conclusion that we should love each other and be more communal and think about the long run and care for our children and love art. In some sense, that’s the generic conclusion of every argument about what you should do because we all just expect people to be trying to argue for that. It’s often funny to see when you go through an analysis, you draw conclusions that aren’t those conclusions.

Jim: Though interestingly in this case, the conclusion looks very much the way people actually live.

Robin: Well, yeah, people aren’t that crazy, but yeah, so it pushes you even more in that direction relative to the ideals you might typically espouse.

Jim: That was a very good introduction. People can check that out on Robin’s essay which as usual will have on the episode page. Let’s start with probably the framing that is most talked about. I’m going to explain why I think Bostrom’s view of ancestor simulation is limited later, but maybe you could lay out his three-part argument a little bit and your take on that.

Robin: Well. Sure. There’s three scenarios, he says, and that they’re the only options and you have to pick between one of the three. One scenario is that as we move on into the future, we will never really be able to create simulations. That is we’ll never get the capability, the technology or maybe it will even die out and we’ll just never be able to create creatures who are living in a simulation and don’t really know that that’s the fact. It requires some technical abilities to do that. You have to be able to create these simulated creatures who have a rich life and who think a lot like a human like us thinks and feels. I have a whole book on what would happen if that technology were to become widespread called The Age of Em, but basically, one scenario says that doesn’t happen. We never are able to make these future creatures.

Robin: A second scenario says we do become able to make these future creatures and we do even make a few, but we don’t make very many. For some reason, when the future is able to make creatures like us who could be fooled into thinking they were living in a simulation, there’s the potential that they could make a lot of them, an enormous number, but they don’t. They just make a few. The third scenario is that the future is able to make these creatures like us and that they do make creatures like us and they make a lot of creatures like us. Therefore, there are more of those creatures in the future who think they are living in a world like us than there are actually people living in the time and manner that we think we live.

Robin: If that’s true, then if you ask, “Well, which of them am I?” you have to conclude you’re more likely to be one of those future creatures than one of the current creatures who thinks you’re in the real world. Most people who think they’re living in our era and in our time are wrong. Most of them are creatures in the future who believe that this is what they’re living. The claim is you have to pick one of these three scenarios. Now, obviously, there could be other reasons why you might be living in a simulation. You might be in some other very different universe which creates a creature like you, but that just adds to the possibility that you might be living in a simulation. The claim is you’ve got to pick one of these three options.

Robin: I pick the option that they don’t care much. I agree that most likely the future will be able to create simulations like us. Most likely, they will create some of them and that will be important, but when I look from our era back and I say the analogies of what we do to make things that are simulations of our ancestors, we make movies, we make plays, we write novels, we play games, or video games, and when we try to simulate our ancestors and do as best we can with that, our interest fades much more quickly than does population. Before 200 years ago, basically, population doubled every 1,000 years, but it seems to me our interest in the past falls more quickly than every doubling a half every 1,000 years.

Robin: Our interest in the year 1500 is far more than twice as much than our interest in the year 500 which is far more than twice as much in our interest in the year minus 500 and so on. That suggests that most interest is concentrated near the people who are able to do the simulations. We are mostly interested in simulating people like us in societies like us. When we go back into more stranger distant cultures, we can do change them to make them more like our world, like say, Game of Thrones does. As you may know, Game of Thrones has characters in cities far larger than any cities that ever existed back then. There are cities that are the size of our cities. Why? Because, well, that’s the kind of cities we live in and that’s what we’d like to see. That’s my take, there’s the three scenarios, you have to pick one. I picked they don’t care.

Jim: Interesting, but I’ll push back a little bit on interest declining very rapidly. If you remember, the very best-selling series, Clan of the Cave Bears, Jean Aurel, something like that, was the author. They’re remarkably fun and they were set-

Robin: As a percentage of all our movies or fiction, it’s really small. I think they did a movie, right? One movie of it?

Jim: I don’t know.

Robin: It was a series. It was three books and one movie. That’s really small fraction of all the fiction we have.

Jim: But the future is a long time. Well, maybe. Of course, we don’t know if the future is a long time. I think some of your other work talks about the great pruning rules in the future and maybe our future ain’t that long. That’s another possibility which is that we may get the capability.

Robin: You should notice there are almost no stories or movies about the other hominids who existed around and near humans.

Jim: That is true. That is-

Robin: Zero of those.

Jim: Hey, you, screenwriters out there. Let’s talk about the interaction between homo sapiens and Neanderthals from the Neanderthal perspective. That would be interesting.

Robin: Exactly.

Jim: Our future descendants don’t care. Let me now open it up a little bit. Again as I mentioned at the beginning, ancestor simulation is only one little wedge of possible ways that we could be living in a simulation. You referenced one is that we could be actually a simulation in a higher order universe, nothing like our current universe. Then for whatever God knows reason literally, maybe some kid playing God on his PlayStation XXXXX has spun up this simulation for entertainment purposes, research purposes, educational purposes, who knows what, maybe ones that we can’t even contemplate. Higher order universe is certainly one.

Jim: In fact, the neat thing about that is it gets rid of some of the constraints in Bostrom’s analysis. He does a fairly good job of convincing me at least that, yes, in the future, it is plausible that we can have the computational power to simulate brains in a jar equal to the number of humans that currently exist. However, if you’re in a higher order universe, you have some higher possible computational capabilities which is enough to essentially calculate the universe itself and simulate that, not just our brains in a jar. Higher order universe opens up a whole bunch of additional things that could be going on in the simulation. It’s not all fake. It could actually be “real” and our science really works.

Jim: Of course, another possibility is that a friend of mine told me about the idea, [Anna Solomon 00:13:37], I don’t know if she coined it or not, she called it the sleazy simulation that even in the higher order universe that computation is going to be limited by economics, as you often point out, think about the economics. Maybe we simulate the earth to 14 decimal points with the quantum mechanics, but maybe yeah, the planets around Alpha Centauri are just wireframes at this point and they only bother to calculate the stuff in fine-grained form once somebody gets out there.

Jim: Then there’s even crazier possible ways to get to a simulation. We’ve talked before or I’ve talked before on the show about the really out-there idea called Boltzmann brains, the idea that if the universe is infinite or effectively infinite, though actually I would say infinite, and behaves in a quantum mechanical fashion the way we believe the universe works, then it’s so exceedingly low probability brains can exist and can pop into existence through quantum fluctuations, perhaps the way the universe was started, that are large enough to simulate our current visible universe, the universe from the Big Bang.

Jim: Here’s the early crazy and sick think about that. If the universe were truly infinite, infinite is a powerful concept, there’s an infinite number of Boltzmann brains large enough to simulate our universe. If that were indeed the case, then it’s almost certain we live in a simulation in a Boltzmann brain.

Robin: Well, no. We have to take the ratio. This is the key point. I’m happy to guess that most likely, sorry, is in fact that the universe is infinite. That seems to me plausible. However then, what we are likely to be in depends on ratios of things in this infinity, and unfortunately, it’s often ambiguous how to calculate ratios when you have infinite universes because when you take the limits in different directions, you get different answers. That’s a fundamental theoretical problem people have in reasoning about probabilities in infinite universes, but we should probably set that one aside here.

Robin: My best understanding is that Boltzmann brain speculation is greatly diminished when you have a proper understanding of how quantum mechanics is supposed to really work. People have talked as if one component of a superposition was a conscious thing and I don’t think that’s right. I think you have to focus on your basically decohered, some parts of the quantum system. That’s why I would actually not focus very much on the Boltzmann brains, but I’m fine with thinking about what if the real universe is different than the universe we see around us. That’s not crazy, right?

Robin: It seems to me the only way to really get a handle on that is to ask, “If we could simulate universes that weren’t the universe we live in, which ones would we simulate and why?” The hope is that at some larger level of aggregation or analysis, the existence of creatures who have reasons for doing things and who compete in some world would be more robust than maybe the details of the physics of the universe they live in. Even if the creatures who simulate us aren’t in the same kind of physical universe as we are, they might still have similar sorts of reasons for simulating something.

Robin: That seems to be our best hope for making a guess about what sort of universe they might live in if they are simulating us and we’re not in this kind of universe that they’re in. If we were going to simulate universes, which we do sometimes they in video games or movies that are different from our universe, what we tend to do is we tend to pick universes which have similar social structures, even if they have some different underlying physical infrastructure and we also tend to pick universes that are less constrained.

Robin: There’s a reason why we love stories with magic. Magic takes away, in our stories at least, many of the constraints that we must suffer under that make our stories harder to tell. Most fantasy is set in worlds which are very different from ours, physically perhaps but socially, they’re much more like our world. The key difference in their physical world tends to be that magic or something like it allows them to do the sort of things we would want to do, except that our physical world makes it hard. That suggests that if there’s a strange universe simulating us, that first their social world is more like ours than their physical world.

Robin: That is they still have creatures who have reasons like we do and perhaps motives like we do which is why they simulated creatures like us with our sorts of reasons and motives and that our physical universe is less constrained than theirs. They probably have to deal with a more difficult universe that gets in the way of their plans even more frustratingly. Those are the two hypotheses I would offer based on the mere fact that we postulate that they live in a very different universe.

Jim: That’s interesting. We could call it perhaps the argument against them such is as far as we know or at least those of us who are ranked naive realist like myself, no magic. No magic is at least a check mark against being in a simulation because if someone were to have simulated it, by your argument, presumably they dialed in something at least, as you said, was more interesting than their own universe. That might well look like magic.

Robin: Right, but our universe does have some amazing capabilities. Maybe those amazing capabilities are magic relative to their world.

Jim: That’s, I suppose, possible if nothing comes to mind. Anything come to mind for you in that category?

Robin: The quantum Hilbert space is to me pretty spectacularly amazing.

Jim: What is the quantum Hilbert space?

Robin: Well, it’s the idea that the physical state of the universe is this very high-dimensional vector space where there’s a linear evolution equation in that very high-dimensional vector space which is how all the universe in some sense and principle can be correlated. Every atom and every spin everywhere in the entire universe in principle can be in these very complicated correlated states and that’s a power that we rarely see being used, but in principle, it’s right there in the math and sometimes we are able to use it.

Jim: We recently had Seth Lloyd on the show and he talked a bit about, “The universe is a quantum computer that is computing itself.”

Robin: Right. The quantum computer that is the universe is a pretty amazing technical machine.

Jim: Here we go. Suppose the higher order universe is strictly Newtonian and they wanted something more interesting, so they dialed in quantum mechanics and general relativity.

Robin: Well, that would be odd because they’re trying to build in their own universe, computers that they can’t build in the universe that supposedly can be built in another universe. That’s a bit dangerous in the sense that if you give your simulated creatures computing powers that you don’t have, then when they try to compute things, you won’t be able to compute them. You’ll have to make up answers and convince them that the fake answers are the real answers and that may not work well if they have more ways to check on the answers.

Jim: Well, let’s add one more fix. It’s a Newtonian metauniverse, but the speed of light is infinite. Therefore, you can build computers and there’s no quantum mechanical effects, so there’s no limit other than perhaps the Planck length at how fine-grained computation can be, and further because you’re not limited by the speed of light, computation can be arbitrarily fast. Maybe something about heat dispersion is different, so you don’t have the overheating problems that you have in our computations. Therefore, they can perhaps throw enough classical computing at the problem to do at least a good enough job that we can’t tell the difference to simulate quantum computing.

Robin: We should probably address this issue people have that many people imagine that if we’re in a simulation, it’s a very faithful, very accurate simulation, and therefore, if their universe doesn’t have the same computing abilities as ours or they’re limited, that there would be errors sometimes and you might be able to tell that you are living in a simulation because of the errors in their calculations. Some people say that, therefore, we’ve looked at some things in our universe with high enough accuracy that they conclude, “Well, we couldn’t be living in a simulation because we would see these errors and we don’t see them.” I think that view is basically mistaken under the presumption that the only way to simulate a universe is to simulate it in high detail and high accuracy.

Jim: I think there’s an important fork here. One is, do you simulate the universe or do you simulate the agents because if you simulate the agents, you can make them think that their …

Robin: Right.

Jim: … experiments are correct to 14 decimal points even when it’s just a shitty wireframe.

Robin: Exactly. Almost all the discussion of simulation is focused on agents and simulating the agents and seeing their experiences, but if you’re just focused on the agents experiences, they don’t experience the universe in fine physical detail. They experience only some broad aggregates of what they see in here. It’s quite possible at a much, much lower cost to give them the impression that they’re in a consistent universe when they’re not in fact.

Jim: Brains in a jar is much easier to simulate than an actual physical universe good to 14 decimal points. I think that’s fair.

Robin: Right. We should just walk through though, just to make clear to people why that’s possible, first of all, if say they create a device and it produces error-prone noisy estimates, those don’t have to be the estimates you show their eyes or their ears. You can clean it up and present them clean data when they see the results of whatever experiments they run. They don’t know that in fact it wasn’t actually run by a real experiment. In addition, if you ever find out that they’ve been able to put together several different things that you didn’t anticipate and showed a contradiction and show that in fact it couldn’t possibly be a real world, well, at that point, you can pause the simulation and reverse it, go back before they started collecting the data that was produced the contradiction and then be more careful and make sure that doesn’t happen.

Robin: If you’re running these simulations, it’s quite possible at relatively low cost to pause them or reverse them, change something and then continue on and make sure you don’t make whatever mistake you are worried about.

Jim: I like that. You could have the metauniversal editor that maintain or the consistency person, a content continuity person on a movie set who make sure that the flowers were read when they reshoot the scene with a vase on the table, not yellow. Of course, there are famously continuity failures in movies and there’s a whole genre of people who look for that, but imagine a continuity editor for our universe that could back out, replace and then replay supposed contradictions discovered by our scientists.

Robin: I want to mention though that I think this topic tends to break people. It’s one of the reasons I’m wary of talking about it a lot or pursuing it a lot. There are a lot of people out there whose minds are not that stable, who can function reasonably well in ordinary environments, but a hypothesis like the simulation argument puts them in some sort of a paranoid state of mind and breaks them. That’s sad because in fact, most likely I would say, we are not living in a simulation. The adjustments you would make for the small chance you live in a simulation tend to be relatively minor. They’re real, but they’re minor.

Robin: Mostly, I think you do okay ignoring the hypothesis that you might live in a simulation and getting really obsessed with hypothesis mostly isn’t very healthy which is sad because I like to be intellectual and explore a wide range of interesting topics and there are interesting things you can learn a bit by thinking about this topic, but it’s a little dangerous.

Jim: I think that could be. People listen to the show know when I talk about Boltzmann brains I usually put out the advice, “Never, never think about Boltzmann brains while tripping on acid.”

Robin: I have this book, The Age of Em and it’s about this future world where brain emulations are the main dominant powers on Earth and in the universe. In that world, they have more uses for simulations than we do. I was able to identify ways in which they might usefully have simulations as part of their regular practices and that makes the simulation hypothesis more interesting or relevant.

Jim: Tell us about that.

Robin: One thing is that sometimes people will have a crisis. For example, one of the reasons we like to watch war movies or war movies about tsunamis or things like that is we don’t experience them very often, but we’d like to be prepared for them. Going through fictional versions of them can let us mentally practice for what we would do if a big rare but important event happened. Well, emulations, basically, they can agree that whoever is running their brain, the infrastructure and the operating system is running their brain, is allowed to put them in simulation sometimes, allowed to put them in crisis simulations. They can do this so that they can practice what to do in a crisis, and so that other people can evaluate how well they might respond to a crisis.

Robin: When you start to see a crisis like maybe we’ve seen over the last year, as an Em, you’d say to yourself, “I might be in a simulation here,” but you would then buckle down and try to deal with the simulation well because you know you’re probably being watched and judged on how well you’ll handle this and you’re going to learn later what you might do better as a result of the simulation and you will not freak out. Emulations, when they see big crisis situations, will be less likely to freak out. They’ll be more likely to just buckle down and do what they need to do to deal with it because they expect they’re more likely to be in a simulation in those scenarios.

Jim: Yet, of course, there’s no need in those [inaudible 00:28:58] but maybe. Let’s think about that. How accurate those simulations have to be, and particularly if you’re an Em who knows your simulation and has been in the side simulations before? It only has to be good enough to be useful, much like agent-based modeling in the quantitative social sciences.

Robin: If you can see clear evidence that it’s a simulation and so can the other people in your simulation, then you’ve moved into the trial practice mode, right? As you know, in a building, often you’ll hear a fire alarm go off, right? The first moment you hear the fire alarm go off, you might think, “Well, this might be a real fire.” At some point, somebody might say, “This is a drill,” but they still might ask you to continue with the drill like carefully march outside, etcetera. There’s a point to having drills where people know it’s a drill, but there’s also points in having drills where you don’t know it’s a drill. I expect they would probably, because it’s relatively low cost, not tell you it’s a drill. They’ll want you to see what it’s like if you’re not sure it’s a drill.

Jim: That’s Interesting and it’s a subclass of simulation that’s less than high fidelity, but the question about whether our universe or what we experience subjectively is a simulation or not, doesn’t appear to be relevant, because at least subjectively, we appear to be living in a, best we know, high-fidelity universe.

Robin: Well, that was the point of my essay on how to live in a simulation. When people talk about living in a simulation, their mind usually defaults to some universe-wide simulation, but my argument is that’s not what you should expect if you are living in a simulation. Most likely, if you are in a simulation, it’s a very small one because those are a lot cheaper and therefore will be a lot more numerous. Most likely, the simulation includes the room you’re in, perhaps a few rooms nearby, a few people in the few rooms nearby you and nothing else. It won’t last very long. It may not last hour, it may not last the day.

Robin: Most simulations would be small things like that even if it looks like you’re in a universe. Of course, when you’re sitting in the room you’re in, you think you’re in a universe that’s all consistent with the room you’re in. You think that there are rooms next door and on and on for hundreds of rooms, but in your simulation, that’s just not true.

Jim: Of course, it would have to have the expense if we’re going to be like ourselves at least, so simulating the memories as well which is not that high, but the amount of bits in an episodic memory is not huge. It adds a lot to the computational load but not a prohibitive amount.

Robin: I’ve never really thought that much about partial brain simulations, but if they’re possible, those I guess would be even more common. I usually think in terms of simulating your entire brain and maybe the entire brains of a few people around you, but honestly, I’ll have to say, well, it could be some situations where we only need to simulate part of your brain to find out what to do in that simulation and the rest of your brain doesn’t actually have to be there.

Jim: I think that could well be reasonable, and if indeed, memory is the most computationally expensive part that can be simulated on the side in a different way.

Robin: Right. As soon as somebody asks what you did last Thursday, then at that moment, the simulation generates that information and pops it in, but until you ask the question, then it’s not there.

Jim: Or it could just be a finite state machine, right? Predefined canned memories that were relatively inexpensive to create. Who knows? Maybe they’re developed by an evolutionary algorithm and they’re running a whole bunch of these things in parallel.

Robin: In The Age of Em book, I noticed a number of different reasons why you might want to a simulation. There’s not just one of them. For example, you might want to simulate to test people’s loyalty to see if in some temptation they would betray you or stay loyal. You might want to see where ideas came from. Intellectual property, of course, is important and good institutions for intellectual property are important. One of the institutions might be that we take the situation where some idea was generated and we simulate it in enough detail to figure out who to give credit for, “What were the real sources of that innovation, that idea?” and simulations could be used for that purpose to.

Robin: If we have like a half dozen or more different reasons we might want to run a simulation, then I realized in fact what we’ll want to do is run simulations which simultaneously achieve as many purposes as possible. You can test someone’s loyalty while you look at where an innovation came from, while there’s a tsunami outside. You can do a number of different things at once. That has an advantage of lowering computational expenses. There’s just a standard thing in experimentation literature about how experiments that do many factors at once or in a sense more efficient than an experiments that test each factor independently and it lowers the chance that somebody in simulation can figure out why it’s there.

Jim: Though to do ABCD testing, you have to run a lot more trials. You have to run the competent torques of how many trials you have to do to get a valid statistical signal.

Robin: Right, but it’s still less than the sum of doing A trials and B trials and C trials and D trials.

Jim: I think that’s probably correct. You say there’s good literature on that?

Robin: Yeah. There is a literature on factorial designs for experiments basically, and yes, basically you learn more about factors A, B, C and D if you put them all together in a single trial …

Jim: Sure.

Robin: … than if you do each one separately.

Jim: Google, of course, does that famously. It has a couple hundred parameters that it varies for every person. Nobody sees the same Google as it turns out.

Robin: Right. Now when you’re in a simulation, even if you have a suspicion you’re in a simulation, you can be much less sure about what it’s for. You can less bias your behavior to make your behavior look good for the purposes of the simulation because there’s multiple purposes and your behavior strategy would be at cross purposes for these different reasons.

Jim: Of course, this same reasoning doesn’t just apply to an Age of Em where the agents are embedded in our known universe and do side simulations for various pragmatic purposes but also apply to the other arguments from metauniverses to Boltzmann brains.

Robin: Right. Again, if whatever universe we are actually in has a similar social world or similar mental world, more similar at least to our social and mental world than its physical world, then they would similarly have these reasons to do cost effective simulations where they achieve several purposes at once with their simulation. Therefore, we can be less sure what the purpose of the simulation we’re in is, especially since they might have other purposes that are uncommon for us.

Jim: Let’s go back a little bit to the metauniverse idea. You talked about some ones that argued from I like their culture, I like our culture, but maybe the simulations from a for a much higher order purpose. One of my favorite physicists who likes to speculate a lot and even publish what are frankly very fancy speculations with some discussion of possible experiments someday is Lee Smolin. He’s got a very interesting hypothesis about evolutionary universes. The detail that he happens to put out is that maybe a universe is the backend of a black hole and that each universe that is created, the backend of a black hole has somewhat of a random resetting of laws of physics relative to its parent universe as it turns out.

Jim: Some laws of physics are more likely to produce black holes than others and so you get this interesting evolutionary dynamic, but if you were to actually have a universe like this, you’d have many, many, many, maybe it’s infinite or maybe it’s a very large number and there’s a big difference between the two, but let’s imagine that the simulation at the metauniverse is at that level. Essentially, they’re doing evolutionary exploration of universe speculation through this Lee Smolin-type application. Then we get no biases about their culture like our culture, our psychology like their psychology.

Robin: Well, that scenario seems to be a scenario where universes are just really cheap. If in fact you can set up some generative process that a universe creates black holes that each of which creates more universes, obviously you’re in a situation where a lot of stuff can happen, very cheap compared to your initial expenses. In a world where computation is extremely cheap, you can just do an extremely lot of it. You may well do computation across a very wide range of possibilities which then of course don’t have to be as near to where you started. Yes, in a world where they are just using crazy wide amounts of computation to explore a crazy wide range of possibilities, we can say much less about their world from what our world looks like.

Jim: I think that’s a reason. I know you often use an economic pruning rule if you assume a metauniverse with no boundaries and the economic pruning rules are very hard to discern.

Robin: Again, when things are cheap, then variety gets large. That’s, of course, the thing we see in our ordinary economic world. When products are expensive, there’s a limited range of them and they’re trying to be more functional. The cheaper product gets, the more variation some of them will have, the less constraint they have by any particular functionality or cost considerations. You can just go while making a wide, wide range.

Jim: It’s entertaining, but it might not be useful.

Robin: Well, it will be optimistic in the sense that, look, all else equal, we should want to live in a universe where there’s lots and lots of creatures out there. It’s really scary to think that we are the only like intelligent life in our visible universe. If we kill ourselves, it’s all gone. Oh, my God. I want there to be a lot more things like us out there, so that if we kill ourselves, other things will go on and there’s just more happy creatures out there.

Jim: We talked about this on one of our earlier podcasts. It’s interesting all good nerdy 13 year olds, I definitely believe that solving the Drake equation said there is lots and lots of intelligent life in the universe, but the more I thought about it, the more I read, the less certain I am by any means including reading your own work.

Robin: Sure, but that scares me. That’s scary to me. If we’re all alone and there’s nothing else in our Hubble volume, but nothing else in 10 to the 20 Hubble volumes around us, then stuff like us is really, really rare and we could really screw it up.

Jim: On the other hand, and I think this is important, and in fact, it might be the most important thing there is, if it turns out that is true and I’m now to the point of, “I can’t decide. Maybe we’re alone, maybe we’re not,” and if we are alone, that gives us a tremendous responsibility and a tremendous source of purpose. People talk about The Meaning Crisis. I say, “Hey, to my mind, until proven otherwise, our meaning, our purpose is to bring the universe to life.

Robin: Overwhelming banquet of meaning, more meaning than we can really eat. Why? Honestly, because if you look at the world, we’re not remotely taking on that task. We are not remotely as acting as if we had such an enormous responsibility. Almost everybody is very locally focused on their own life and their own little considerations. Indirectly, it turns out favorably, the economy and society is growing, and we have prospects to grow a long way, but we’re not really organized and focused on making sure that happens. We are stumbling along.

Jim: But maybe we should be. I think that’s my point, right? If we were to stop and think about this, why I keep saying that the Fermi paradox is the second most important question in science, the first one being, “Why is there something, not nothing?” but are there others out there or not has gigantic implications on everything humans do. It seemed to me to take the precautionary principle somewhat seriously, we ought to act as if we’re alone until we prove otherwise. It may be that we prove otherwise fairly soon or maybe we never prove otherwise.

Robin: Ideally, you would think we just have some scale or coordination parameter where if we don’t need to coordinate very much, we each just do our own thing, and when coordination gets more important, we would turn the dial up and make more personal sacrifices in order to coordinate better and then we would coordinate better. That would be the thing that a rational creature would do, right? When you look at what happens to humans, when we turn that dial, really crazy scary things happen. When humans believe that they really strongly need to coordinate, they fight wars, they create totalitarian governments, they do lots of crazy stuff that’s pretty disruptive. That scares me, I say.

Robin: Well, if we’re in a world where we really do need to coordinate a lot and then we tell everybody, “Hey, everybody, stop being so self-centered and let’s focus on coordinating because we have this huge responsibility,” and I look what people have done in those situations in the past, I go, “Oh, my God. Maybe we’d be better off not being told even at the cost of a big chance of failure.” I don’t know, but either direction is scary, having this huge responsibility and failing to meet it is scary and putting together the usual organizations and coordination that people have in the past when they thought they are dealing with really important big things, that’s also scary.

Jim: That’s scary, but it may be necessary, learning how to actually cooperate at very large scale without putting our civilization at risk every time we do it.

Robin: That’s where I would put my number three questions along your list. First question, why is there anything? Second, is there anybody else out there? Number three, how do we coordinate?

Jim: I would agree. Do we operate cooperatively at scale, at the giga scale or the tera scale?

Robin: I say we know tentatively the answer which is we have lots of hypotheses for better ways to coordinate and we should be testing. We should be doing many small scale trials of various ideas we have for better ways to coordinate. We are failing even to do that. That’s the thing I most want to push people to do in the world is to try small scale trials of our many innovative ideas for a better way to coordinate.

Jim: Interestingly, that’s what our Game B Movement is all about, not all about, that’s one of the main things. One of the things I’ve been involved with for the last nine years is defining what some of those better ways to operate might be. We’re finally, finally getting to the point where 2021 we’re actually going to start trying.

Robin: Jim, I will praise you as being one of the people who maybe takes Principle Zero which is the most important thing is to think about what the most important thing is and then do stuff related to what you think the most important thing is. The vast majority people don’t do that. It’s very rare to find people who actually think about, “Gee, what’s the most important thing in the world?” and then focus their energies on doing that thing.

Jim: Unfortunately, we were designed or evolved, I should say, “Don’t get rid of that design.” We were evolved to live in hunter gatherer groups of 150 or less. It had some loose affiliations with maybe 1,000 or 2,000 people and we had low energy source, fire being our most energetic source and not too much ability to meddle with the big picture of the universe, but all that’s not true anymore. I think the necessity for focusing on-

Robin: I’ve just got an error here. It said, “Quota exceeded error. You’ve run out of free disk space to save a local backup of this data.”

Jim: Let’s wrap up here, just in case we’re losing what we’re getting out here, which is at the end of the day, what do you think about? If you had to put a meta-bet on in your opinion marketplace in the metauniverse, do we live in the various kinds of possible simulation? What are your thoughts?

Robin: I still got to put it on a few percent. Well, a few percent living in a kind of simulation where the physical world around me is not really what I think it is. Now, if we are in a universe that was created for some purpose and you want to call that a simulation, I think that’s much more likely. I think all bets are off on basically where our universe came from. It could well have come from some intelligent source, produced with some program in mind, but that’s the scenario where basically the universe as we see is the universe roughly we see. It’s just has a purpose and a source. If we’re saying the universe around me is not what I think it looks like, that’s all an image just to make my brain thinks it’s there, I’ve got to put that and say no more than a few percent.

Jim: I think I’m with you that I can’t even get any principle way to think about the metauniverse problem. It’s just like above my pay grade and humanity’s pay grade at the moment, but of course, it’s quite interesting that the history of humanity is taken up with nothing much else other than speculating about that kind of stuff. Nonetheless, we don’t really have any way to have any principled idea on why the universe exists in the form that it does.

Robin: I guess I’ll push back and I’ll say, “Look, if you’ve tried for a while and you haven’t found a method, a way to do it, then it’s fine to quit. I don’t think you can conclude that we don’t have any way to do it. We just haven’t found one yet.”

Jim: It’s not even obvious how we would.

Robin: Right, but that doesn’t mean there isn’t a way. It just means we’ve tried, we’re giving up now and somebody else can try if they want.

Jim: Exactly. In terms of things like Boltzmann brains or other artifacts of infinite universes, I’ve chosen principally on aesthetic grounds to reject infinite universes, again when it comes to metaphysical speculations where there is no evidence one way or the other-

Robin: What I would say there is evidence in how we do indexical reasoning, but that would take us on a whole different conversation. There’s a whole literature on the indexical probability and lexical priors. They do suggest, in my opinion, that infinite universes are pretty likely.

Jim: I’d love to do that sometime but not today. Anyway, I personally reject infinite universes to get rid of weird artifacts like Boltzmann brains and other oddities.

Robin: Of course, many people that reject the simulation argument on the same principle, “Hey, that’s weird. I don’t want to think about where universes.”

Jim: Or the other one, Occam’s Razor which we know is not a law. It’s just a heuristic, which is-

Robin: It’s a pretty conservative heuristic. It just says, “I don’t want to think about weird shit.”

Jim: Or not a little stronger than that, which is that really complicated shit is probably less common than less complicated shit.

Robin: But the main argument is because, “I don’t want to think about it. It’s a problem to think about complicated stuff. I’m going to stick with the simple thing that I can think about more easily.”

Jim: That goes to your point that it may be unhealthy not to have some level of simplification if we’re going to live one’s life, right? “If I’m going to spend all my time worrying about whether we’re inside of a Boltzmann brain. What the fuck that means? I wouldn’t get anything done.”

Robin: Absolutely.

Jim: All righty. Well, any final thoughts?

Robin: It’s been great talking again and I’m going to hope for a number five.

Jim: All right, it’s been it’s been fun.

Robin: Bye.

Production services and audio editing by Jared Janes Consulting, Music by Tom Muller at modernspacemusic.com.