Latest / Art of the Question / Debunking the Simulation Hypothesis - Prof. Franco Vazza - #40

Transcript

Debunking the Simulation Hypothesis - Prof. Franco Vazza - #40 · Art of the Question

• Julius Moorman: Franco, you have such an interesting profession. To people who don't know what you do, how would you describe it? So I am an astrophysicist. I teach at the University of Bologna. Yeah, I do research on how the universe works. Then this said, astronomers, astrophysicists, they can specialize on very different scales. You can go from the solar system, how the sun works, how other stars work. I work on the ⁓ largest scales one can imagine. So I focus on how essentially the largest structures existing in the universe form. We call them clusters of galaxies, then which are places in which galaxies orbit together around the common center of mass. And we're speaking of hundreds of galaxies. And then those clusters of galaxies, they are interconnected through filaments and all these structures all together we call the cosmic web. And each clusters of galaxies is several, let's say up to 10 million light years across. Every single cluster of galaxies and the entire cosmic network in principle can be infinite. We don't know if the universe actually is infinite or finite and repeating itself in some periodic way in a sense, meaning it's a closed surface, it's a closed four-dimensional surface that one can cross many, many times, but the scales we can see are organized on scales of billions of light-years. And in particular, out of these giant assembly of structures, I'm particularly interested in how magnetic fields are organized. how sometimes some structures light up in the radio wavelength, which means they are accelerating relativistic particles and so on. And since, ⁓ so if you do astrophysics on scales of planets or stars, you have the hope of seeing some of these systems evolving in real life. More or less, stars can explode from one year to another, planets, of course, they orbit many times, et cetera. So you have some sort of, you can see how they evolve at least across the lifetime of an astronomer or you can still use ⁓ real observation, maybe from your predecessors, still in modern time to build up an history of how a specific system can evolve. The stuff I'm studying is that is evolving on scales of hundreds of million years. So there is no evolution we can see, we can... look at the same object, at the same cluster of galaxies all our lives, actually all the extent of humanity can point at the same cluster of galaxies, still you wouldn't see anything moving on the telescope, which means what you should do is to build a fictionary or plausible evolution sequence using computer models. So you teach a computer everything you think you know about physics, meaning Newtonian gravity, hydrodynamics, general relativity and so on. You combine all the different bits of observations that we do have because maybe I should elaborate a bit on that, but if we can observe very far with telescopes, we are also accessing different slices in time of the evolution of the universe because light, even if it's the fastest thing that exists in nature, still is a finite propagation speed is 300,000 km per second. So if we see something which is when we say a billion light years away, it means lights took one million years, exactly one million years is almost a redundant definition to cross that length. Meaning we cannot see how that particular galaxy is now. We can only see how that galaxy was one million years in the past. And actually this exercise goes on until 13 billion years in the past. And so depending on the wavelength, depending on the instrument, we have different sampling of different epochs in the universe. Of course, not for the entire celestial sphere, only for the beats we can sample here and there. And so we computer models and using physics, we try to make consistent histories that can connect together all these different pieces. So we see certain galaxies when they were just born, and then we see very evolved structures. And we got an idea pen and paper. At the zero level, one can get this idea pen and paper of how cosmic structures evolve. But then if you want to compare in detail and really test models and understand if your particular model is working or not, you should rely on simulations. because then you can also have mock observations of how the universe should look like. Then you compare them with images you have already or that the next generation of telescope is going to produce. And so you need the short story that we need very sophisticated computer models and very big simulations because we try to reconstruct an holistic story of several processes and pretty much this is what they do and many, and many other people are doing. It's incredibly interesting. And so basically, if I understand it correctly, we have to use simulations because of the mere skill of things. If we get our telescopes and we look at some planets that are pretty close. can actually see them moving on a time scale that is relevant to humans. But we are talking about such enormous structures that it seems to be still, but of course everything is moving in the universe. And that's why these simulations can help us see how the universe is constantly evolving and also where we came from. Yeah. It's a combination of these things are extremely far actually in astrophysics. We can only do lab experiments. Basically, we have sample of rocks from the moon. If we are lucky, we get samples from asteroids, meteorites, stuff that is impacting Earth constantly. We can catch some very energetic particles called cosmic rays. Now people are sending ⁓ spacecraft on comets and bringing back some fragments, but I mean... Everything we do in astronomy is not really lab work. So the closest to have an experiment in which you do test hypothesis, you predict the outcomes, and then you see if this is consistent reality or not, are computer version of laboratory experiments. And our data are everything that we can get through photons. And this is, I mean, everything we are studying is not accessible. with any direct measurement, direct meaning taking samples like you would do in biology. The second thing is the time span, yes, which is just a sort of prospecting effect. you see stars, so planets, we do see them moving because they are very nearby. Then you start putting these structures far and far away. All stars for centuries, they seem to us pretty much still. Now we can have telescope observations. We go to very high resolution, we do see that those stars are moving. You need, of course, to increase your resolution as much as possible to see the tiny, tiny movement of stars in our own galaxy. And sometimes if a star is exploding, you can even see that with naked eye or with amateur telescopes. But then everything happening outside of our own galaxy is not moving. It will forever remain, say, in the same pixel in the sky because It is so far that in order to cross the physical extent of that pixel, it will take literally millions and millions of years. The universe is constantly evolving, but not on human time scales. The closest we can see to something extra galactic changing is if we see in a distant star, in a distant galaxy, a particular star lighting up that would be a supernova. This is still something we can see, well, not really with naked eye. but with small telescope you can buy for your own recreational time. This is still accessible, but this is not a motion. This is a light ramping up in a galaxy and then getting dimmer on a time scale of a few days. Everything else on human scale appears to be perfectly still because it changes on time scales at least a million years. Yeah, now that makes sense. And when did you first become interested in running simulations and the field in general and what sparked your love for this field of work? So when I was a child, I wanted to do an archaeologist and then by random readings coming probably from philosophy, et cetera, I got also interested into cosmology. So existential questions related to what's going to happen to the universe, what's going to happen to planet Earth, et cetera. Then I discovered the most powerful tool for trying to address this question probably was physics. So I enrolled in astrophysics. And then among the different fascinating things, I encountered the concept of turbulence, which is something, is one of the most ubiquitous phenomena in the universe. I'm really speaking of the same turbulence that is mixing coffee and milk. And the fascinating thing is that using different physical processes, different chemical compositions, going from simple milk to plasma to ⁓ planetary atmosphere and so on, is the same underlying mechanism repeating itself on many, different scales. So if you understand that and if you study it, all of a sudden you have a very powerful handle to understand many things in the universe. ⁓ And this is very fascinating, but also we are into the regime of chaos. cannot, even the most powerful computer on earth, even in the generations to come, will be unable to solve exactly what's going to happen in your coffee and milk mixing together. You can only get a statistical solution to the problem. You know how energy is going to flow, but you cannot predict where exactly one parcel of milk is going to end up after you stir it with a spoon. Because that's chaos. It's almost to the best we can know in physics. This is a sort of ⁓ a unbeatable barrier in uncertainty. You have super tiny fluctuation that really can arise from the scale of even quantum fluctuation. Of course, we think they don't matter, but if you accumulate the perturbation after another, you can have a vortex in your milk appearing on a different part after a given evolutionary time. The same problem we have in astrophysics and If you want to have a statistical prediction of this to some, with some level of realism, you need to use a numerical simulation. Numerical simulations are employed when turbulent fluids are involved. If you want to design the stability of an aircraft, if you want to model the the innermost part of a nuclear power plant, of a tokamak, if you want to... improved engineering of pipes. Whenever there is turbulence, you want either to minimize it or to boost it because it depends on the application. If you want to combustion, you want to enhance turbulence. If you want to have engineering application, usually you want to reduce turbulence. In all ways, in all cases, you go through a different sort of numerical simulation. So I got to learn about numerical simulation and then the connection was how can I study the largest turbulent phenomena in the universe? And they ended up studying cluster of galaxies and so on. that was more or less the connection. And once you enter into chaos and turbulence, you don't go, I mean, it's so fascinating. You find connection all the time at those scales. And so you stay there usually. Yeah, it's really interesting. And what I'm curious about is Compared to when you started running simulations and to today basically, how much has the technology evolved in those years and how much better is it today than it used to be? ⁓ It has improved a lot, but still not enough in the sense that when I started my thesis, we are speaking of 20-ish, a bit more years ago, I was given to analyze simulations that my supervisor produced using his own workstation below his desk, basically. Something you can run on a single node. Now we are using ⁓ parallel computing. I'm daily in accessing, for example, the largest computing center we have here, which used to be, and I'm not checking the ranking every month, but last year it was the fourth on the planet. And the simulation is employing hundreds of thousands of cores. So it's a massively parallel simulation. You want to distribute your simulated problem into as many cores as possible. All these, the kind of stuff I simulating is highly parallel because you, I mean, to speak very simply, you have a huge cosmic volume. You divide it in a regular lattice. Every piece of the lattice is being evolved by a different core, like really a CPU or better a GPU, a graphic processing unit nowadays. This is solving a small part of the problem. And then you assemble everything in a time step and then you update your computation and then you proceed to the next time step for tens of thousands of time step. This sort of simulation are very popular in astrophysics because you can imagine you can assign the evolution of a single galaxy to a computer. And then you can evolve at the same time the evolution on another galaxy on the other side of your simulated universe. The two galaxies are almost independent, one with the other. So the computation is heavily parallel, but at some point you want to update their position, how their mass is evolving, how the gravity is changing. So, and this is progressing steadily in time, almost steadily. There is a famous Moore laws that is telling how, what is the speed at which the speed of processors is increasing in time. So there is some regularity in that. But still it's not that we are doing revolutionarily different simulation than the one I was first given 20 years ago. We are adding more physics. It's not that the volume has increased dramatically. It's that we can put more physics because observations are forcing us to introduce more physics. By more physics I mean Maybe 20 years ago it was enough to decide, this star is going to explode or not. It's plausible that this star in the simulation is going to explode like a supernova or not. And you just put all the energy around. Nowadays, since we have spectroscopy, we know the elements, we know the velocities of the material, etc. You have more complex models, which means your algorithm has more functions, more lookup table, more everything to decide, OK, that star is going to explode like that kind of supernova and not the other, meaning he's going to have a particular yield of chemical elements onto the surrounding elements. And these elements are going to have their own nuclear reactions and blah, blah. So you are adding more, we would call that sub-grid physics, stuff that is happening on scales you cannot resolve directly because we can really not resolve all astronomical processes from. planet formations to clusters of galaxies in the same simulation. So at some point you decide everything that happens below a given pixel size or resolution element, I cannot directly resolve, but I'm going to account for its physical evolution in some approximated way. So this is as greatly improved into the last 20 years because observations provided us more clues, more puzzles, more information. The size of the universe we can simulate a steady increase. But I mean, when you compare with something that is potentially infinite, even if you double the size, still you are comparing with something. So in that sense, it's not that we can... I mean, the increasing size is not the game changer here because the universe really seems like a repeating pattern after some particular scale. So usually you put all your numerical efforts going smaller in scales rather than bigger and bigger. Right. what, yeah, of course the topic of today's discussion or one of the main topic I would say is the simulation hypothesis. ⁓ There are some different definitions out there. I was curious to hear your definition of it and perhaps also if there are ⁓ some other ones that people use often. Sorry, definition of simulation. Yeah. So how do you define the simulation hypothesis? So when people are speaking about whether we live in a simulated universe or not, how do you define that problem? Okay. Yeah. That's an interesting question because, yeah, it can give you two different answer and the answer you find can be more or less satisfactory depending on what was the initial point. ⁓ So I think we can start with probably the article that started, it was not really, really the first article, but this is the one that around which like interest cluster, which was an article from 2003 by philosopher Nick Bostrom, Bostrom, Swedish philosopher. And the title it was, are you living in a computer simulation? So I think that I'm not a philosopher, but okay, I got interested into the issue. That was the techno version of a long standing puzzle, which is there in philosophy since forever. There are several philosophers. There was a famous ⁓ Chinese one that now I don't remember the name. Sorry. I can name Rene Descartes, example, ⁓ George Berkeley. So several philosophers that at some point coming from different perspective, they ask themselves the question, How can I know if what I think is the reality? It's a creation of my mind, some sort of dream, some sort of hallucination. It may be the creation of someone else's mind. And the most recent take, starting from 2003 probably, is what if this is instead the result of a computer simulation? Now, defining what a simulation is, okay, it's interesting here. Coming from physics, a simulation for me is a physical model in which you teach an algorithm some rules and the rules are continuously repeated and they form some sort of ⁓ digital reality. In our case, mean, in the simulations I do, there is a main code. The code is just learning, not learning, is given a set of rules. to put things in a simple way really is the law of gravity, the fact that the universe is expanding at the rate that we measure with telescopes. And if you want to make your simulation more complex, you add hydrodynamics because we have normal matter colliding, et cetera, and say chemistry and Maxwell law for electromagnetism. You just teach these rules and there is, okay, people took... several decades also to understand how to solve these equations mathematically, having as few errors as possible because, I mean, it's not easy to solve an equation by iterations, etc. etc. And then you just let the simulation go. With some initial conditions that in ⁓ cosmology are a very smooth and uniform background of density and temperature fluctuation. Then the simulation is evolving. And from that point on, you don't do anything else. The simulation is marching exactly based on its own rule. And you can basically do the same for other systems. You can do that for planetary systems. You can do that for just a single galaxy in which your initial condition would be, for example, an already formed galaxy spinning. And your task is to see if if the set of stars is evolving in a realistic way, et cetera, et cetera. Now. Since this team, I think, has become very popular at the same time in which also, of course, computer games became increasingly more popular and more charming. I'm also, I mean, I like to play video games myself. Now, the two things have been a little bit mixed together because you call simulation also something that for me is a video game, meaning something that I don't think has a simple set of rules, but also has many additional information. And the whole goal of ⁓ a computer game is to fool the player to not to fool entirely, but to make the player, I mean, to give the player some sort of illusion of reality. So there is not really a physical model only, but there are tricks to make the simulation or the video game faster. Like you don't care about simulating the entirety of your universe only where your observer is looking in that moment. The two things are pretty much mixed together, I guess. So when I try to tackle this question, I had in mind the simulation in the sense of the physical simulation. Other people had more in mind probably the video game side of things. ⁓ When this article became popular, it was also exactly almost when the Matrix, the movie was released. And it was not the first time in a science fiction story there was this concept. Of course, ⁓ I guess, Philip K. Dick had plenty of books about that and many others. But okay, it was a very nice movie and the technology also movie making was mature enough that it was, I also remember it, I got really impressed. And so the two things became popular altogether. So can computer mimic reality? Can computers from other civilization mimic our own reality? How can we know if we are simulated by? So I think in the way I read, this paper, are you living in a computer simulation by Boston in the way that Matrix, the movie was made? I think we can do spoilers of movies of now 20 years ago. Right? Yeah, yeah, we can. Yeah. Okay. The idea was that we are being simulated by our future meaning we think we are living in this Apple, but actually the real time in some real clock is a couple of hundred years in the future. And for different reasons, our descendants, which in the case of Matrix actually are machines that took over the control planet Earth, are simulating the past. So they are trying to mimic how the Earth was 200 years in the past. In the case of Matrix, I don't fully remember the plot, but it was to make ⁓ people like ⁓ quiet, I don't know, because that was the Earth just a bit before... this big revolution of ⁓ machine domination, whatever. So the idea is that we are still part of the same reality with the same identical set of rules. It's just the future that want to simulate the past, which is not super different from what I actually do. I try to simulate the past evolution of this universe, except on scales incredibly larger. People are simulating the possible past formation of the planet. ⁓ Also to study, for example, the evolution of climate or you can simulate the past evolution of the solar system to understand the deformation scenario of the moon and stuff like that. I mean, it's not a crazy idea. We are not just there yet at the point of simulating the past of civilization. But then you have video games that are exactly doing that. You have video games that try to simulate civilization from the start to the end. So the two things are mixed together. I focus on the idea, can we be part of a physical simulation in which there are set of rules, a song computer is evolving and whose purpose is not to fool each of us or just me because I can only be sure that I exist as a consciousness. But rather my question is, can there be an holistic physical simulation, which is not the goal of fooling each of these individuals, but is trying to reconstruct a holistic global view of the world or of the universe. And we happen to have, I mean, have been, to have arose within the simulation as a result of chemical processes, et cetera, et cetera. Yeah. And I know you are not necessarily a believer in the simulation hypothesis for reasons we will get into. ⁓ more in, if we look at the practical aspect of things, but I think what is interesting to kind of. Yeah. Think about is the following. Usually the argument I hear for the simulation hypothesis is a logical one. Right. So there are a few options. ⁓ one option is that we live in a society where we will never figure out how these simulations work. So therefore ours must be the real one. Another one is that we will eventually figure it out as a society, but we will just, we won't run one of these simulations that is so similar to our own. Or there is another one, which we will get so advanced that we will be able to run very hyper-realistic simulation, it's very similar or exactly the same as to what we're experiencing right now. And if we figure that out, we will probably not run one simulation, but we will run millions of it. And in that case, if the latter one is true, the odds of us living in one of those simulations is very high and the odds of us living in the only real one. are very low. That's my very basic understanding of the philosophical logical argument. Is that more or less correct or did I butcher this? From the logical point of view, guess that no, that's sort of correct. Yes. think that's a fair take of the main argument by Bonestrom and which has been refined over time. Yes. ⁓ I mean, when I first heard of that, I was a bit puzzled, but okay, fair enough. I I think you can build the same argument thinking of, I don't our synthetic drugs will continuously become more and more sophisticated. They will constantly give ⁓ humans a more refined and powerful experience of reality. And in the future, there will be many more of those and humankind is going to probably grow in So maybe we are not in a computer simulation. We are in someone else's hallucinations for the few. I think you can repeat the argument with many different variations invoking dreams, hallucinations, simulations. I don't know. I was not super interested into that, but I found it okay. A fascinating take. Then at some point, maybe it took me way too much time, but I realized, oh wait, but. Simulating universes is exactly literally what I do every day, except at the scales at which we can at the moment. And I also had in mind, okay, every time I run a simulation, either it's on my laptop or it's somewhere else, there is heat coming out of the computer. I know how much energy I'm burning when I'm doing my big simulation on the computing center because they want you to know in terms of kilojoules, etc. And so every simulation is costing some energy. So whatever is the logical or philosophical take, there is energy attached to that. So you cannot rescale. Well, now I know, but before I wasn't sure, but You cannot just rescale the simulation idea at Libitum without not having to pay some bill. And so the task was, okay, let's assume different scenarios. Either we simulate the entire universe or we simulate just planet Earth because we are part of a simulation whose only goal was to simulate planet Earth and civilization and us. What's going to be the cost of that in terms of energy? And I was not concerned, of course, with electricity bill, the current cost of power, etc. I was using fundamental quantities like how much, what is the minimal amount of energy you should burn, even if you are able really to convert mass into energy like Einstein prescribed. So what if you had the maximally efficient conversion of mass into energy? How much of that do you need? And then we can start to use numbers. And if you're using already the lower limit, if the lower limit makes no sense, it means that in reality, the real energy consumption is going to be a thousand times more. And if the lower limit already does not make sense, from my point of view, you can be sure that simulation cannot exist. Okay. In this universe, that's exactly, this is where a different take on the simulation hypothesis are going to different in in different directions. So what I tried to do in an article that I published ⁓ last year, I guess, yes, was can I test how big, how complex can be a simulation in this universe? Because I know the rules of this universe. Those I know. What I cannot say is what if we are simulated by a different universe with different physical laws, with different Dimensions, et cetera. Maybe energy in those universe will cost less. Maybe going to higher dimensions, things are easier. And for example, at this point, I'm curious to think even about that, this we cannot debate so much about that because we don't know the set of rules. I think we can debate of the ⁓ ancestors hypothesis, which is also the same one to which the original article by Bostrom is alluding to, in which the Matrix, the movie is taking place, or can we be a simulation by our future descendants wanting to simulate their ancestors using the same playbook of this universe? My answer would be no, at the moment. I mean, based on everything I did, I would say absolutely not, it's impossible, because it will require an amount of energy which is not impossible because it's too expensive. It requires to burn like entire galaxies, which is something you cannot do. So that's my take on this. If you say, yes, but we are actually being simulated by a different universe, by different creatures in that universe, the laws are different. This is outside the... the boundaries of science at the moment. don't see how we can even tackle that hypothesis. Yeah. And how do you even go about ⁓ calculating something so complex? Because at the beginning of our discussions, you already mentioned that it would be impossible to simulate ⁓ milk being stirred inside of a cup of coffee. So it gets really complex, really fast. ⁓ How do you go about estimating the amount of energy it takes to run a simulation of ⁓ this size? ⁓ just let me clarify. It's impossible in the sense that you cannot predict exactly what's going to happen. And therefore this is exactly what we do with weather forecast. It's impossible to know how the weather is going to look like tomorrow on a scale of hundreds of meters. So what we do is that we collect the initial conditions for today, meaning we collect all satellite measurements and maps, et cetera. We evolve many, many, many models, meaning we simulate the same exact scenario with different small perturbations. And then we see the probabilistic outcome. When the weather forecast is telling us there is 80 % chance of tomorrow raining, it depends on how they gather the information, but this is... Lose related to the fact that if you run a hundred times the same simulation with small variations, since we are dealing with a chaotic system, 80 % of times you get a shower at that point. And in 20%, maybe the shower is happening tomorrow or it has dissolved, something like that. So it's not that it's impossible to run the simulation, but it's impossible to be sure of the prediction. can only be probabilistic. but this said, this is a deterministic system. And you know exactly if you decide what is the minimum scale I want to resolve. So I know I need to assign a float, a given ⁓ precision in how I describe numbers to every single resolution element of my simulation. I know what's the volume I need to sample. I divide it for the number of resolution element. I know what's the total memory I need even to start the simulation. And then I know or I have an idea of how many time steps, what is the time scale I want to resolve? Do I want to resolve the time at which molecules are hitting each other or at which an electron is orbiting around the atom or quarks within the atom are evolving? It depends. This is putting more and more more stringent constraints on the time stepping of the simulation. But once I decide all these numbers, I just can measure the finite amount of memory and also computing power, meaning flops, floating point operation per second, I need to evolve. Then, since moving any bit anywhere in the universe, it can be in a machine, it can be on any other kind of support, moving a single bit costs energy, is not, even if we think we type stuff on a keyboard, we do stuff on our mobile, Maybe we don't perceive the energy which is used, but anytime we are flipping a bit, there is an irreducible amount of energy that is being used. And this also physics has established since a long time. So we don't know exactly what is the minimum amount of energy that we need to flip a bit. Okay. That depends on the temperature. We know the minimum one, can have the lowest possible temperature in the universe. If we adjust the temperature, we know what's the cost of that. And so we can just multiply the previous estimate number of, I mean, the memory amount which is needed for that energy. And we get a total amount of energy. We need to write things somewhere and then we can update our estimate for how frequently must our system be updated. That in a nutshell is how you compute memory and energy requirement. And, ⁓ and then it depends. So For example, in the article I wrote, there were different scenarios. One is, what if I want to simulate the entire universe down to the smallest scale that physics is telling us that exists, which is called the Planck length. It's an incredibly small scale. It's 10 to the minus 35 meters. Let's say there are 60 orders of magnitude, so 10. And then you have 60 zero after the first 10. of difference in scale between the largest scales in the universe and that smaller scales. So just for the sake of the argument, you can compute what if I want to simulate the entire universe at that resolution, how much should that cost? And you get an energy which is like 30 times 30 or the zone magnitude larger than what actually is in the universe is extremely more expensive to run the simulation than to have the system itself. If you want to simulate even the tiniest bit of its spatial description. This can, I know it can sound a bit weird, maybe an example is maybe everybody knows or many people know that the three body problem is unsolvable. There were also the famous three body problem, et cetera. which has to do with the fact that describing the orbits of one planet around the dominant star is an easy, easy, into quotes problem because it took Newton to solve it. If you are the third body, unless the body is a perturber, so it has a very small mass, you have three bodies. A three-body problem, to be solved in a perfect way, requires basically infinite calculations, but The physical system of three particles with three points requires incredibly small... So, if you want to create three electrons, say, orbiting... Well, let's say one proton, one electron, and another electron orbiting around each other, it takes an extremely tiny amount of energy to create that system. If you want to simulate that system, even supposing it's Newtonian gravity, which is not, but just for the sake of the argument. It will take infinite amount of computation to get to the real solution because that problem is unsolvable, meaning you should continuously refine your computation forever. So creating the thing and simulating it can also take incredibly different amount of energies. And so the simulation of the entire universe is just impossible, but this was like a trivial computation. Okay. And then you can say, yeah, but no one proposed that. Okay. Fair enough. Then you can. shrink all of a sudden the size of your problem to the size of planet Earth. Because now we are discussing the case in which we believe or we acknowledge the fact that every human being is real and the consciousness of every human being is real. And now what does it take to have all these people living in this system? And okay, in the article I was saying, okay, let's pretend all astronauts, so people that actually went far beyond the boundary of planet Earth were like, non-playable characters in video game, like those 50, 100 people have been put there just to fool the rest of us, but okay. And all the rest we see outside of planet Earth is like, I don't know, a screen or some continuous feed of fake information just to keep astrophysicists like myself busy all the time. What would it take to simulate the entire planet down to Planck scale? And again, you get an incredibly large number. Now, I don't remember. Let me go back one second to my note. So that was also orders and orders of magnitude. Yeah, impossible. If you want to simulate only planet Earth, so you shrink the size of your actual volume from the entire universe to a single planet, still the amount of memory you need is equal to basically taking an entire galaxy and burning it. all together, so a system made of 100 billion stars, burning all these stars together in a nuclear flash, and take this energy just to write everything on the disk. And then you start the simulation. What I try to stress in any possible way, this is not a technological limitation, it's just a limitation of physics. It's like saying, yeah, I can speed up my simulation going faster than light. No, you cannot. There are hard boundaries dictated by physics. ⁓ before I forget, that's actually a good analogy. If you take Bostrom's argument, which I think is a good argument. I don't want to make fun of it. It's a very legitimate philosophical argument. At some point, physics is entering a discussion. If you take the same mood, you would say, yeah, okay, human beings first started ⁓ walking on their feet and then they invented, I don't know, the wheel, the bicycle, Formula One cars, the airplane, spacecraft. So, it's natural they will go faster than the speed of light. No, it's not. This logical progression, if you believe physics is right, and if you don't, you should show it is not right, there is a hard barrier which is dictated by the speed of light. no matter what your logical extrapolation would suggest. Same things, ⁓ a bit more complicated here, has to do with computation and simulation. There is a hard barrier on how much energy you should pack somewhere in the universe. So our descendants in this hypothesis should pack all this energy somewhere to make the simulation sound convincing. So, simulating the entire planet is impossible. They should pack up an entire galaxy in a small, I don't know, on a size of maybe a super giant planet, burn all that energy, converting all the energy of a hundred billion stars just to ride the simulation somewhere and then continuously burn the same amount of stars. So they should take another galaxy. So now you have the business is to move galaxies around to continuously evolve the simulation. That's impossible. No matter what our philosophy is saying. Then the most interesting last part was, okay, but no one is really perceiving the Planck length. This is a scale that is there in physics, but we cannot directly access it. It's nothing observable even remotely. We're speaking of 10 to the minus 35 meters. It's far below the scale of the innermost part of quartz, et cetera. And so I assume the smallest scale that we can directly measure, that we can really see in our experiments. And the nutshell is that we can use neutrinos. Neutrinos are these particles that we do measure. Neutrinos have a wide spectrum of energy, but the one I was using here are the ones coming from extra-galactic space. they can be, I mean, everything that is happening outside of the sphere of our simulated planet in this hypothesis can be fake, but we know these particles are crossing our planet and actually they... innermost part of our planet and because they are detected after they are crossing the entirety of our planet and they emerge on the other side and we have particle detectors and we know these neutrinos are continuously crossing the planet and they can come from every direction. So in that sense, they do a tomography of the planet around every possible angle. And so we can catch this particle. We know what is the wavelength associated to them. That's okay. A bit more into quantum mechanics, but if you know the energy of a particle, you know how compact is the portion of space in which this particle is compressed. I'm really speaking in a super loose sense, but in physics you can associate the energy and the scale at which a particle is fluctuating because deep down we are not dealing with like hard spheres, we are dealing with fluctuations in space-time. Okay. So we know what that scale is. We can know the detection of neutrino for knowing what is that scale. And that scale is 10 to the minus 16 centimeters. So we are considering something which is 20 orders of magnitude larger in size than the plant length. So the simulation, of course, is becoming cheaper and cheaper in absolute terms, but still So again, doing all the computation, what if the entire planet is simulated only down to this small scale and everything else is just something to keep up busy. Still, we get an incredible amount of energy, which it is the power that our sun is emitting isotropically in two minutes, which on astronomical units is very small, to be honest, but okay, technologically, It means we are capturing the entire radiation of our planet isotropically, so somehow we trap it in all directions and we use it to run the simulation down somewhere, the initial condition of the simulation down somewhere. This in principle could be done, but we are not still done. The next step is that if you want to evolve the simulation constantly, the amount of energy required is again Impossible. You have to take a supermassive black hole and burn it continuously every second forever. So again, now we are down to taking supermassive black hole, are conglomeration of 10 to the 9, sorry, a billion solar masses, a billion solar masses, burn them continuously just to march the simulation. Okay. So. Again, is not in that case, this is not impossible in absolute terms because the universe, unlike the other two scenarios, the universe has that energy somewhere, except now our future descendants somehow have to take not entire galaxies, just the innermost part of galaxies. have to move constantly all these black holes, burn them somehow. I don't know how you burn a black hole, but that's the requirement. And that energy is just enough if you have 100 % efficiency to make your computer go. And I didn't even tackle the fact that of course all this would heat up something. And so all this incredible amount of heat you have to release somewhere. I I stopped there. It's impossible for the universe. I think it's still impossible for every possible civilization, but okay. And this is where I stop basically. So for me, now I feel more reassured. Let's put it in that way. Yeah, no, I understand. And one argument, it's not my argument, but when I hear people talk about the simulation hypothesis, ⁓ people also talk about rendering similar to video games like we discussed in order to preserve ⁓ memory within the simulation. This is more for the audience, but you know how in Grand Theft Auto if you go into a super fast car or an airplane, if you fly super fast, you'll see that part of the map isn't rendered yet because the game will only render what the player is seeing and is loading in the map as you go. because this preserves a lot of memory and especially on the older consoles this was necessary because they went very powerful so they needed to be very smart in how to use all that computing power. Now some people argue that we don't need to simulate the entire universe or the entire earth right up until the Planck scale all at once but that perhaps the conscious being living on that planet solely what is necessary to render their conscious experience would be enough. Maybe this is a different version of the simulation hypothesis that could very well be, but I'm interested to hear your take on it. No, no, I'm familiar with that because after I put this article, it was published in the peer review, then there were a lot of commentaries and I was of course very curious. I really take issue and got offended when I started reading, ⁓ but this person has never played video games. That really takes issue because I do play and enjoy video games like Elden Ring, etc. So I know how video game works. I know all the tricks or some of them. Yes. Now, the things are interesting. There are two things I want to discuss. The first thing that no one is discussing, which is probably the biggest part of it, is that no one has the remote idea or understanding of how you create consciousness if it is a digital process at all. or not by saying it is not a digital process and not the I'm not meaning in any way it is a supernatural, but it can be just a biological process. No one can show no one has ever showed that it is just the result of computing in the sense that yes, the brain metaphorically can work as a computer, but okay, architecturally the way memory store, et cetera, is radically different. But there are plenty of theories that says Maybe consciousness is not exactly arising from the computation. Maybe intelligence is, as computers are intelligent, but consciousness, so this incredible and elusive fact that somehow there is something perceiving all this happening, can also be the result, for example, of electromagnetic fields that we know are in the brain superposing one to another. So it can be a superposition of electromagnetic fields. We do know electromagnetic fields are superimposing. Even if no one has the faintest idea of how that can happen, other people are going more towards biological processes. The key difference would be, for example, again, not my really core expertise, but of course I was reading and thinking a lot. So just to understand the difference of the two, if consciousness is just a digital process, you can take the same architecture. I mean, you take the architecture of a computer simulating a consciousness, you can assemble it as you want. You can put it in a cube, on a plane, in threads, whatever. The consciousness will still be there because the algorithm is the same. So that would be, that would mean consciousness is fully a digital process. It could be, okay. If it's not a digital process, but a physical process, that would mean that if you could ever take like the neural system of a brain, preserve all the computation, but assemble it in a different way, I don't know, spreading it on a surface. In that case, the consciousness would be gone because the consciousness in that case would be the superposition of electromagnetic fields, which are topologically in a given portion of space, exactly as you can have different sort of inter-references as you assemble a set of antennas in a different way. This would be still physics, nothing supernatural, but would be not a digital process anymore. Or if you change the same elements which are doing the computation into from, I don't know, silicon to something, a biological tissue, for example, according to other explanation, the consciousness would be gone, but the computation will be still there. It will still be there. Okay, so I mean, we have not the faintest idea and also more intriguingly, It's very difficult to be convinced of either of these hypotheses because I mean, unless it's you exactly there in that machine or in that thing simulating a consciousness, I mean, that's a really hard, is what David Chalmers is going, the hard problem of consciousness. It's very, very difficult to understand how this is emerging. But this said, when we speak of the simulation hypothesis, the simulation is fooling us, et cetera, we are giving for granted probably the biggest I mean, we assume already the solution to the biggest problem we have in physics. So that's a bit taking a big leap. We just assume the collection of computers can create consciousness. And really we have no idea. There is no proof of that. It can be a different process since ⁓ still a physical one. Okay. This said, coming back to the video game things. In that case, we are down to looking for artifacts and it's plenty of artifacts, even in the best computer simulations we have. If you are a single player, I mean, it's a game you want to relax, fully, I mean, relax into this thing. When I play Elden Ring, I pretend I am on my horse or whatever. But of course, if I pay attention, I see plenty of artifacts, artifacts really not on the plank length on Okay, on the size of macroscopic things, you can argue, okay, the artifacts are going to be on smaller and smaller smaller scales because the simulations is getting better and better. Okay, I get that. Then the question is, what's the smallest scale? The smallest case probably is the one at which I don't see any artifacts. And sadly, there is physics and physics has A lot of businesses measuring things at super, super, super small scales. And the smallest scale is the one I quoted is the one probed by neutrinos. Of course, it's not that each of us has neutrino detectors in our pockets, but still there are plenty of microscopical phenomena. So I find that argument childish in the way it is posed because in a video game, the only thing you want to fool is basically the eye of who's playing. But there are plenty of things that are happening in our world from the propagation of electromagnetic waves, the way in which lasers are working, the way in which computers are working that are requiring processes of quantum mechanics that are taking place constantly on incredibly small scales. The fact that we can speak across the globe, that we can use GPS is using physical processes taking place on incredibly small fraction of time. So your simulation cannot run in a fraction of a second, in a millisecond. You need to go to incredibly smaller time scales to make all the reality we see around us consistent. And again, in the paper, I do acknowledge that the task of fooling one conscious being, now it's impossible, but you can foresee that happening maybe in a couple of hundred years, because if you just want to feed a single brain with enough information to make the experience of this unique conscious being rich enough to think there is an entire universe outside of it, that you can. Okay. The problems arise when you try to have a consistent experience for many, many individuals, because then they start comparing cross-correlating information. If I am getting a physical reading here of some experiment and my colleagues are getting the same physical reading in Australia, South Africa, in Cambodia, wherever, you cannot fool each of us with a different set of a rendered low resolution simulation. We are speaking of potentially billions of people having same experiences and of these, some millions of those people are having the same physical lab experiment, lab reading, lab measurements. Okay. So at some point, in order to have a consistent model of that, you just have to build exactly the simulation we are speaking about. Okay, if if my colleague here is observing an astrophysical phenomena at a given coordinate on a time scale of 10 to the minus 10 seconds, and we are synchronizing our clocks across the globe, and we do see the same experiment, and that source is sending a high-energy neutrino, and both of us are seeing the same event, we have the same physical model down to the same super small scale in time and space. And if many people are getting the same, you need a consistent model and then the limits are the one I try to compute. And we are not, of course, yet there the fact that each of us has a neutrino detector, but we all use GPS synchronization. We all use transmission of electromagnetic waves. We all use ⁓ computers, we all meaning the quantum processes inside computer. So the amount of synchronization you need to make this a consistent experience is exactly undergoing the limitation as to fulfill the limitation I pointed out. ⁓ so that's the constraint, I guess. Yeah, no, I, I understand. And I, I'm, I'm also curious. to hear apart from the ⁓ logic, the physics, is there ever even a way to figure out whether we live in a simulation? Can we ever have conclusive proof or is it by definition impossible to ever come to a conclusive answer? So, ⁓ if you want to specify in this question a little bit further, I guess, yes. I would say if I have to bet money, You can exclude we are simulated by our future descendants, not because of technologies, but just because of the rules of this universe. There are plenty of things we still don't know about the universe, that's true, but all the limitations I quoted are coming from the most fundamental one with all the R and we think are really true, like speed of light, Isenberg and certain principle. Those are the most well-tested predictions ever in physics. And quantum mechanics is the most well-tested theory we have. So all the limitations I quoted basically are coming from this. What we cannot exclude using physics, I guess it's we are simulated, meaning a single conscious being that in this case will be me. But in your case, if you exist, it's going to be you. And for every people in the postcard is going to be themselves. This we cannot exclude, but we are not in that case very far from what already Rene Descartes was having a big trouble describing centuries ago. His solution was there must be God giving the property of existing to things, just super-simplifying. Because there's a fundamental gap between the subjective experience and everything we can do in physics, which is objective. No one has come up so far with a way to bridge this huge gap in scale, which is the same we need to understand conscious a little bit better. It's something that only one person at a time can feel as far as we can tell. And convincing the other that you are conscious or convincing yourself that anybody else is conscious, apparently is not something that you can easily do using the physics tool we have. But okay, you can find some correlates, some proxies for that. I hope someone is getting a physical solution of that. ⁓ The other way out, and I don't think it can be easily tested, but now I'm curious, I will. This is a sort of research I do in summer or over Christmas time, but let's see for the next summer what's happening. The second big escape into the simulation hypothesis is that we are getting simulated by an entirely different reality. To explain things in a simple way in this paper, I call that the Pacman hypothesis. So if a character in Pacman at some point somehow is becoming so intelligent to realize that they are in a simulation, because maybe he's detecting artifacts. Then maybe he's really seeing the pixelization in his own experience. Then the next task is to guess what's outside of the simulation. And that I think is going to be impossible because in Pacman you cannot even think gravity exists because there is no gravity. Maybe you do all your computation in 2D because you're living in a 2D world. Maybe you do a lot of computation using how they are called. You don't measure energy in arcs or joules. You measure them in power pellets because they're the only energy amount you do know your simulation. So in that case, it's impossible to go out of the simulation. The best you can do probably is to understand there are artifacts, which is what maybe we can still understand. I if we find something spectacularly off in the sky, like, I don't know. the, an, an, how do you say, a non number in the universe and not a number in the universe or some big mistake or some big, I don't know, some big commercial playing out on the scale of the Milky Way or something like that. Yes. Okay. That's a bit big artifacts and then we can think something is off. Maybe that's the best we can do, but then to, to, to even guess what's outside and to put limits out of that. If the physics is different, we cannot do, ⁓ we need to know that playbook. So at that point, the question is, why do you really, really, really think we are in a simulation? I would understand if you, yeah, to me it's weird at this point to get there in the absence of a blatant artifact. So if I do see something that really doesn't match up with everything else we know in a truly spectacular way. Then I would understand. I understand philosophically, it's a very nice tool to make you think about, what do I really know? Where, which are the gaps in my knowledge, et cetera, et cetera. But I don't know. I wouldn't, I wouldn't go there at this point. I don't see really the logical need beside some philosophical exercise. No. And I think you touched on something interesting there and it One of the things I wanted you to ask ⁓ as perhaps a final question of today's discussion. I find it incredibly interesting to think about this stuff. You wrote papers about it. Bostrom wrote a paper about it. I believe he even replied to your recent article recently and also put your paper or his answer to your paper in a frequently asked question. It's got an extremely popular in social media, but in scientific discussions as well. is it even, should we even care? Like does it even matter if we, if we live in a simulation or not? Because either way, doesn't really impact our life. ⁓ I mean, it's interesting as a, a philosophical exercise. It's interesting as a physics exercise. ⁓ but do you think people should care about this? Does it even matter? I'm very curious to hear your take on it. Well, I'm not qualified to answer. ⁓ mean, I think it matters as much as it can matter to, I mean, the answer you give to the question ⁓ is my life having a meaning? Is there a grand design? Is just a random fluctuation of things? I don't know. So, ⁓ you can say that the answer, mean, an answer that we are in the simulation for some people can be equally disturbing as we are part of a universe which is like fully deterministic and there is no free will, which is one possible interpretation of physics. It can be equally disturbing or it can be less disturbing, meaning, okay, maybe I can change things somehow, or it can be, it can replace The idea that there is a God and you should have a religious take on life. I think he's getting on the same basket of problems and answer. ⁓ I'm pretty much relaxed about that, to be honest. ⁓ Maybe if one is asking the, if one is getting the obsession of this question, okay, it's signaling that the meaning you give to your life would change a lot if you know it has been already designed by someone or no, or you are concerned if someone is watching you somewhere out in the real world or not. I am not, but I understand if people are. Sadly, so I was a bit surprised, I must confess, when I look at some take on my article, when it was put on some popular media, also on the Italian side, et cetera, I was shocked that I think several people, at least those that are leaving comments online, which are not actually sampling. mean, they're not a fair share of real people, but let's they really, really wanted to have the simulation hypothesis remain. And they were offended or annoyed by the fact that someone was trying to question that this was a bit surprising to me, but okay. Fair enough. learned something. Yeah. ⁓ is there, is there anything, ⁓ you think? We haven't touched on or anything you would like to emphasize before we wrap up today's discussion. ⁓ well, that's another personal take. can, I can mention that to me, the connection to this problem besides, so what made me realize I had some pieces that they couldn't contribute to this discussion, was that when I was asked to buy, by my, my former Astro amateur society in my own town to give a public ⁓ conference. I didn't want to say the usual galaxies are big, the universe, blah, blah, blah. At the same time, there was yet another take by Elon Musk somewhere. I guess I still was on Twitter at the time before I left. And it was yet again, one of these things, ⁓ yeah, maybe we are in a simulation just because he likes that. And so I was very annoyed that He didn't feel any urge to elaborate more on that. it goes at an extreme. mean, you can see those tides that are coming over the web when some of these people are speaking about that and then people just follow. And he also found extremely concerning when I found another interview of ⁓ Elon Musk and the reporter asked if you can like If you have the most super powerful computer at your own disposal, what's the first thing you would ask? And allegedly his question on camera was, what's outside of the simulation? And I found, I cannot even quantify how much depressing and disappointing and refining I found that one of the most powerful person on earth, not for the good reason, in my opinion. would burn his question to ask something so silly, not tell me how to cure cancer or to fix climate change or whatever. He was burning his question for something so stupid. So let me think a lot of what can be in the mindset of some people like that. And sometime when you hear this simulation hypothesis coming in tides, it's coming from people that have some interest, at least that's my interpretation, they have some interest in making you more and more interested into what's a second life, meta, into confusing what's digital and what is real. So sometimes I also think maybe it's unconscious, but the two things start to mix together. If you say, there is really no way to say if we are in a simulation or not, maybe we are. And so who cares if we are, I don't know. polluting because it's all bites anyway, who cares? Something like that. was really annoyed by all this take, some take of the problem. Yeah. Yeah. I can imagine. can imagine. ⁓ Frank, this was so incredibly interesting. I've learned a lot and it was, yeah, it was great. I think it's very important to have some nuanced takes out there ⁓ that aren't just like following headlines then I think you do a great job at ⁓ trying to explain with physics why the simulation hypothesis is a bit more complicated than many people make it seem to be. So thank you for that. Where can people go if they want to keep up with the work you're doing, the papers you're publishing or perhaps, ⁓ yeah, some other channels that you're active on? ⁓ I don't want to be disappointed, but that was really a one-shot work. do 99 % of my research is in boring astrophysics, like boring, I don't know, black holes, colliding clusters of galaxies and stuff like that. And I am not on the most popular media, which is is Mastodon. So if you want, yes, I am Mastodon or otherwise I don't do I mean, I don't plan to do any more article on this unless sometime you get the idea. Maybe you can debunk or test this or this other part of the hypothesis. So no, sadly, I will remain still in astrophysics for for a long while, I guess. Yeah, but I mean the work we're doing in astrophysics is everything but boring. no, that's clear. If you want me to put something in the show notes, just let me know. For now, ⁓ yeah, for now, just thank you so much for talking today. I had a great time and yeah, hopefully we'll speak again someday. Okay. Thanks a lot for having me.