Do We Live In The Real Universe?

Nuclear war, out of control artificial intelligence, planet decimating pandemics, asteroid strikes, superpowered gamma ray bursts. There are many ways humanity could meet its end, and philosopher Nick Bostrom has been trying to list them all. Bostrom spends most nights alone in his office, working obsessively until the early morning. When he does sleep, he uses a bed near his desk. He almost always seems concerned, and maybe with good reason, for he has found one particular threat to humanity that is seemingly unavoidable. In 2003, Bostrom wrote a provocative paper with a potentially dire outlook for the human race that stated that either we all live in a simulation or we're all doomed. The argument goes like this. To begin with, assume that intelligence and consciousness don't require a biological brain. That it might be possible to create a machine intelligence with hardware and silicon rather than neurons and sodium ions. Assume also that as time goes on, our computing prowess grows ever more sophisticated and that someday, far into the future, we could build a computer so powerful that it is capable of simulating the entire universe, up to and including the appearance of digital beings that are fully conscious, curious, and aware of their surroundings, but completely unaware that they live in a digital recreation. How many digital beings would exist? The number could be enormous. Today, there are roughly 8 billion people on Earth, which represents roughly 7% of all the humans that have ever lived since we first arose to consciousness 200,000 years ago. But compare that to the trillions upon trillions of digital creations that we have made in only half a century. And so once some intelligent species somewhere in the universe, whether us or somebody else, successfully builds a digital recreation of the cosmos, Bostrom argues there's no turning back. The number of simulated conscious beings will inevitably dwarf the number of organic brains. And by sheer weight of numbers, the chances that we are a simulation become nearly a certainty. If it's possible to simulate the universe, the odds say that we should be living in a simulation. That is, unless there's something preventing the appearance of simulated beings. Bostrom argued that there was another possibility, a way out of the inevitable conclusion that we almost certainly live in a simulation. It could be impossible to build simulated universes with simulated brains because no civilization becomes advanced enough to accomplish it. That all intelligent species reduce themselves to ashes once they harness the energies and technologies capable of such a feat. In other words, we can rest assured that we are definitely not simulated if the end of our species is right around the corner. And so if we could conclusively determine whether or not we live in a simulation, then we could know what fate is in store for humanity. But how could we do this? Is there a way to figure out whether we live in a simulation or not? Is there a way to save the human race? [Sponsor segment: Nomad eSIM.] How much information would you need to simulate the entire universe? The summit of Cerro Pachón in the dry, high altitude Coquimbo region of northern Chile is a lonely place. It's surrounded by a vast desert, with what little scrub and succulents there are largely supported by occasional mists that blow in from the coast. Some animals make a meager living here, hunted by foxes and migratory birds. And yet this isolated mountain serves as humanity's focal point in its attempts to peer into the cosmos. The Vera C. Rubin Observatory is not a normal telescope. At its heart sits a digital camera the size of a car with a resolution of 3,200 megapixels. And every single night, the telescope scans nearly the entire southern sky, producing roughly 20 terabytes of data. This is so much that astronomers had to build an entire dedicated digital infrastructure to pipe the data down from the mountaintop to a processing facility in La Serena, almost 100 km away. Indeed, during its planned 10 year mission, it will accumulate a total of 500 petabytes of data, equivalent to every single thing written by every single human in all of history. It will map over 17 billion stars in the Milky Way and 20 billion galaxies in the local universe. And it will do it again and again and again, night after night, providing an unprecedented time lapse of all the activity in the heavens. However, the Vera Rubin Observatory will barely scratch the surface of the sky. Its catalog of stars will represent less than 5% of all the stars in our galaxy. Its catalog of galaxies, the largest we've ever produced, will still be less than 1% of all the galaxies in the observable universe. For every supernova it records, it will miss thousands. And this is a pattern repeated across telescopes and experiments throughout all of human history. We've barely scratched the surface of the cosmos. All 93 billion light years across we have access to, and all 10 to the power of 88 particles it contains. Indeed, in a 2009 paper, physicists Charles Egan and Charles Lineweaver calculated that our universe has a total information content of roughly 10 to the power of 122 bits. A single star has 10 to the power of 59. A gram of water, 10 to the power of 24. A rough estimate of all the data contained by all the computers and all the servers on Earth comes in at roughly 200 zettabytes. That is also 10 to the power of 24. The universe is very big. And even that is just the observable universe, the only part of the cosmos that we could ever access or have any causal contact with due to the limited speed of light in the 13.8 billion years since it began. Beyond that boundary, who knows? It could be infinite. And yet, despite the relative paucity of information, we have learned much. We can trace the history of the universe all the way back to some of the earliest moments of the big bang. We can reconstruct the formation of the solar system and the lineage of life on Earth. We understand the mechanisms of the cell and the interaction of its molecules. We have traced the rules of quantum mechanics and the subatomic world. And so if we had the capacity, do we know enough about how the universe works to simulate everything in it? And so perhaps here lies the key. We have to work out if it's possible to create a one to one simulation of our universe, and hopefully through that process discover whether Bostrom's fears are founded or if humanity can be saved. Let's start with the cosmic and work our way down. One of the preeminent software tools for simulating the entire universe, at least at large scales, is called AREPO. First released in 2019, it continues to be updated and maintained today. The code starts by dividing up a portion of the universe, a volume of up to a few billion light years on a side, into a large number of individual cells. For the very largest simulations, individual cells can be much smaller than galaxies, but this requires using some of the largest, most powerful computers on the planet. Each cell keeps track of what the universe is like inside its volume: how much dark matter there is, how much regular gas, the temperature and pressure of that gas, any magnetic or electric fields, the state of any radiation, and so on. It's as if you took a chunk of a galaxy, smoothed it over, and described that chunk by its averages. The code can also estimate how many stars are inside that cell, if there are any supernovae occurring, and if giant black holes happen to be belching out matter and radiation. Given a state of the universe, the code can then leap forward in time using all the appropriate laws of physics to figure out what happens in the intervening millions of years. Dark matter shuttling around due to its gravity, gas collapsing and igniting into stars, galaxies approaching each other and merging together. And AREPO adapts, automatically assigning more and smaller subdivisions in the complex parts and freeing up memory in the empty parts. But for all their impressive power, cosmological simulations cannot simultaneously capture both the evolution of the universe and, for example, the dynamics of a single planet. For that we must turn to another kind of simulation. On August the 13th, 1958, a team of physicists and meteorologists at the University of Chicago created the first hurricane crystal ball. At the time, tropical storm Becky had formed in the Mid-Atlantic and was steadily marching westwards to the Americas. The team used digitalized data of the storm's current status, its wind speeds, and pressure readings throughout its volume. They then fed this into a physical model that predicted how the storm would advance. And in doing so, they created the first ever computerized hurricane prediction. And today, government agencies around the world employ dozens of monitoring satellites to provide real time data, which feeds into giant computer simulations that run around the clock. This means that when a large storm approaches, we can finally speak with remarkable confidence about its future track, guided by our simulations of the future of the world. But yet again, these powerful simulations and exquisite predictions still have their limitations. Their view of a major storm is limited to a resolution of around 1 to 2 km. Below that, they must average it out. In other words, they can't follow or predict individual raindrops or the molecules inside the drops themselves. To do that takes yet another kind of simulation, one powered by PlayStations. It is 3 minutes past 3 on the 23rd of September, 1992 at the Nevada test site in the southern United States, and all is eerily silent. The hot sun beams down on the cracked desert ground as a collection of small white buildings absorb and reflect its brutal midday glare. Then, suddenly, as the clock ticks over 1 second to four minutes past 3, the ground itself begins to violently shake. Something has exploded deep below for the last time. This was the Divider nuclear test carried out by the Los Alamos laboratory, 47 years after they carried out the very first, the 1,054th nuclear test carried out by the United States. It would be the last by either the US or the Soviet Union before the signing of the nuclear test ban treaty in 1996. And indeed, ever since that nuclear test ban treaty, the United States has not directly tested any of the weapons in its nuclear arsenal. And so how could it ensure the reliability and performance of those bombs should the need arise? The answer: simulate their inner workings as they lie in storage. This has made the US Department of Energy one of the biggest proponents of large scale supercomputers. And yet the road to the most powerful computer in the world had an unlikely waypoint. In 2010, the US Air Force Research Laboratory wanted to build a new supercomputer of their own. And looking around for vendors to supply the chips for the computer, they realized there was already a commercially available option. In 2006, Sony had released the PlayStation 3. They had collaborated with IBM to create the Cell chip, a powerful microprocessor. And it turned out that the kinds of computations that the Cell chip excelled at were very similar to the kinds of calculations that the Air Force needed. And so in 2010, researchers with the Air Force plugged together 1,760 PlayStation 3 consoles and got them to work together, creating the Condor Cluster, using that machinery to process high resolution satellite imagery for only a tenth of the cost of a typical supercomputer. Today, our simulations of various physical processes span almost the entire breadth of physical existence. They can take us from the earliest moments of the Big Bang to the present day. But to go even deeper into the bedrock of reality, we must cross into a completely different realm. We must cross into the quantum world. Could we find computational clues hiding in the fabric of the universe? If we were to go looking for the fingerprints of a creator or a systems architect, we might not find them in ancient texts or burning bushes, but in the very constraints of physics. Every programmer knows that even the most powerful hardware has a budget. You can't render an infinite amount of detail at an infinite speed; the system would crash. And indeed, when we look at our universe, we see analogies to precisely these kinds of optimization tricks. Take, to begin with, the speed of light. To us, 299,792,458 m/s feels like the cosmic speed limit, a barrier that nothing can cross. But in the language of computation, the speed of light begins to look a lot less like a law of physics and more like a global clock speed. After all, it is the maximum rate at which information can be updated from one cell of the universe to the next. If the supposed simulation is to remain coherent, no signal can travel faster than the hardware can refresh the frame. Then there's the fine tuning problem. Why is our universe so perfectly calibrated for life? If the strength of gravity or the mass of an electron were even slightly different, stars wouldn't form and we wouldn't exist. To the devout, this might suggest a benevolent creator. But to a programmer, it suggests parameter tuning. The simulation that we lived in might be version 1,000,001, one of the few successful runs where the variables were balanced just right to see what intelligence would do. And finally, there is the Planck scale to consider. As we zoom into the fabric of reality, we eventually reach a point where the smooth curves of space and time break down into a tangled mess. And so some theories of physics like loop quantum gravity envision spacetime as discrete, almost pixelated chunks, just as a digital image dissolves into squares when you lean in too close. The entire universe could have a minimum resolution. And so perhaps what we see as inviolable constraints on physical reality are just system settings designed to maintain efficiency. Maybe the universe is trying to save on memory. Indeed, by only rendering the subatomic world when somebody is looking at it, the hallmark of quantum mechanics, the computer that runs the universe could save massive amounts of processing power. But other than that, quantum mechanics would perhaps be the biggest obstacle one would face when creating a one to one simulation of our universe. A seemingly completely unpredictable subatomic realm where objects can be in two places at once, and nothing is ever truly certain. Even the most modern computer would collapse attempting to recreate its various intricacies, and thus the macroscopic world that forms from it. That is, of course, unless the computer itself was quantum. It was the early 1980s and a young David Deutsch was at a party hosted by legendary physicist John Wheeler. Ever the gregarious host, Wheeler was entertaining his guests with expansive speculations about the fundamental nature of reality, and indeed had thrown the party specifically to encourage conversations between guests about the topic of computing. There was an idea circulating in physics circles at the time, one heavily promoted by Richard Feynman that Wheeler had become interested in. If the fundamental nature of reality is inherently quantum, then how could classical computers ever simulate reality at its deepest level? Feynman said at the close of a 1981 lecture, "Nature isn't classical, dammit. And if you want to make a simulation of nature, you'd better make it quantum mechanical." And yet, Feynman and Wheeler, for all their formidable intellect, knew how to pose the question, but didn't know how to answer it. Deutsch, however, was different. And just a few years later he would invent the quantum computer. The problem facing physicists was that as their sophistication in computing grew, they faced insurmountable obstacles when trying to simulate quantum systems. Subatomic systems were just too complex, and capturing the vagaries of quantum mechanics pushed traditional computers well beyond their limits. Ironically, software could simulate the formation of a galaxy or the dying moments of a star, but a collection of simple subatomic particle interactions was beyond it. And at the root of the problem was a fundamental incompatibility. Classical computers, as in every single computer used in everyday life, are built on ones and zeros. A binary mathematical language that reduces all information into exceedingly long strings of on or off states. This mathematical language then maps onto the states of transistors in a circuit. They're either filled with electrons or they aren't. But binary on or off languages just don't make sense in quantum mechanics. And so this would be a hugely important hurdle to overcome in creating a replica of reality, for any super advanced civilization to truly create a functioning simulacra of our universe in all its detail, or any other universe that follows similar rules. In the subatomic world, nothing in our classical experience makes sense. Particles can exist in multiple states at once. You can't know the outcome of an experiment until it's over. Everything is probabilistic, and particles don't act alone. They can entangle with others so that the behavior of one affects the others, no matter how distant. All of this meant that Deutsch would have to work out something new. And to best understand Deutsch's approach to quantum computers, imagine you have a rodent problem. To solve the problem of a mouse in your home, you decide to buy a cat. You set the cat loose and it wanders around from room to room. However, you are impatient and you decide to speed up the process. You buy more cats, one for each room in your house. But this solution isn't perfect. You'll be stuck with taking care of a lot of extra cats. And so what if there was another, quantum way to catch the mouse? Instead of the usual pet store, you go to Schrodinger's House of Kittens and buy yourself a quantum cat. You take it home, place it in a single room, and head to bed, hoping that this new expensive cat does the job. And indeed, in the middle of the night, you awaken to a noise. Getting up, you find your quantum cat sitting calmly by your bedside. Then you walk into the hall and there is an exact copy of your quantum cat, and another in the bathroom, another in the kitchen. Copies seem to have spread throughout every room in the house. You decide to go back to bed. And so it is only when you wake up in the morning that you find a single quantum cat next to the sofa in the living room with a dead mouse at its feet. The quantum cat took advantage of superposition, the ability of quantum systems to exist in multiple states at once. Instead of having a lone cat prowl from room to room, a single quantum cat simultaneously exists in all possible rooms. And when its job is complete, it resolves to the solution: a single dead mouse in a single room. To put this idea into practice, quantum computers rely on something called qubits. Qubits represent the fundamental unit of quantum computation. Unlike a traditional computer bit, which can either be on or off, a qubit exists in a superposition of both on and off states. And so by very carefully manipulating tiny quantum systems and keeping them stable, like the spin of an electron or the polarization of light, these qubits can be both on and off simultaneously. Instead of a single thread of execution winding its way through a problem, a quantum computer can exist in all possible solutions at once. Through careful manipulation of the entire quantum system, the incorrect answers fade away, and the computer's end state represents the correct solution. But for Deutsch, this wasn't just a breakthrough in computing. It was a portal to another universe. He remains an ardent defender of the many worlds interpretation of quantum mechanics, which states that different universes branch off every time a quantum probability arises, with these different universes achieving different results. "The quantum theory of parallel universes is not the problem, it is the solution," he wrote. "It's not some troublesome optional interpretation emerging from arcane theoretical considerations. It is the explanation, the only one that is tenable, of a remarkable and counterintuitive reality." And so for Deutsch, quantum computers harness the power of parallel realities to rapidly speed up computation. And while others disagree, and nobody can definitively agree on an interpretation of quantum mechanics, nobody doubts the computational power of quantum systems. Today, the most powerful quantum computer in the world, at least that's publicly known, is IBM's Condor machine. It contains 1,121 qubits. Of course, that is trillions of times smaller than the processor in your phone. But because quantum systems are inherently powerful, it can still solve certain kinds of problems much more efficiently than any other computer. And though the computer itself is small, its support structure is gigantic. The processor is surrounded by an elaborate steampunk chandelier of cooling tubes and wires known as Goldeneye. However, quantum computers aren't a solution to all computing problems. Indeed, they only excel in certain special cases. And if it wasn't for the fact that one of these special cases is a cornerstone of the modern digital world, the encryption that secures internet communication, there would most likely still today be just another arcane study in theory. But there is one more thing that quantum computers can do that classical computers cannot. They can easily simulate the quantum world, because of course they are the quantum world. "The most important application of quantum computing in the future is likely to be a computer simulation of quantum systems," Deutsch has written, "because that's an application where we know for sure that quantum systems in general could not be efficiently simulated on a classical computer." Quantum computers are a macroscopic manifestation of subatomic systems. Classical computers struggle enormously when dealing with quantum mechanics because they require so much memory and so much processing time to keep track of all the variables associated with even a handful of subatomic particles. But quantum computers are designed to handle exactly these kinds of problems. Indeed, in a sense, the fact that they have real world applications is just a fun bonus, not the reason Feynman and Wheeler first suggested them. And so, any kind of total universe simulator will require a quantum component. When a star burns hydrogen to helium, it is quantum mechanics that dictates the rates of reactions. When light filters into the chloroplast of a plant cell, it is quantum mechanics that determines how much energy is absorbed. And even when we speculate whether we live in a simulation or not, it is quantum mechanical interactions that power our thoughts. We live in a quantum universe. There is one further barrier to our full universe simulations, and that's the idea of emergence. Just because we have physical laws that describe phenomena at certain scales doesn't mean that we can use those laws to tell us about other scales. For example, we know that quantum fields provide the backbone for force and particle interactions. And we can use quantum field theory to describe how atoms work and what goes on in our particle accelerators. But we have no quantum field description for how a protein folds or how neurons communicate or how stars evolve. We have to turn to other theories, other laws of physics that apply in their respective domains. Indeed, so many aspects of nature emerge from collective action that no fundamental theory can capture. So many aspects of nature emerge out of the mist. And there is one key emergent feature of the universe that challenges us most of all. We are sentient beings. We are conscious. But we are not separate from the cosmos. We are a part of it. And so whatever machine that can compute the bits of the universe must also be able to replicate our own emergent thought patterns. When a group of engineers at Google's DeepMind division released a paper in 2017, they had no idea that they would be kickstarting a revolution in computing and igniting a furious debate about the nature of consciousness. Their landmark paper, "Attention Is All You Need," is now considered a watershed moment in artificial intelligence. In their paper, the Google engineers devised a brutally simple scheme that allowed a neural network, a software replication of neurons connected together, to capture the essential meaning of a text by jumping from word to word, weaving together their shared concepts. This engine of attention gave rise to the explosive growth of large language models and the ability of computers to understand the meaning of natural language and generate long coherent messages back to us immediately. Some AI proponents proclaimed that we were currently on the cusp of artificial general intelligence with all the broad general capabilities that a human mind has. Though many, however, disagreed. But what is not under debate is that large language models certainly don't learn in the same way we do and don't process or store information in the same way as a human brain. But if we are the end result of such an attention engine, billions of digital neurons firing in a vast silicon matrix, what would that mean for us? Would that take away our agency, our free will? In a traditional classical deterministic universe, we're often viewed as biological machines, our fates determined by the microscopic collisions of atoms. But of course in a simulation the stakes are different. If our minds are code, then our thoughts are algorithms. In such a world, free will might be nothing more than a feature of the software, a feeling of agency designed to make the simulation more realistic or more productive for the observer. And yet there is still a glimmer of hope. Many simulations that we run require a random number generator to prevent the created world from becoming a static, predictable loop. In our own universe, that randomness is found in the quantum foam. And so, if the architect of a universal simulation uses true quantum randomness to drive the simulation, then our futures might not be pre-written. We may be living in a program, but we would be a program with open ended, undetermined variables. And so, we've reached the end of our current computing limits. We can simulate stars and planets and weather and some aspects of life. And with further advances in quantum computing, we may be able to go deeper into the subatomic realm than our current technology allows. As far as we can see, with enough energy and ingenuity, we arguably could create something similar to our existence, to our universe. But of course, were we to do so, the question would remain: what would that mean? What would be the consequences of a simulated reality? Once Zhuangzi dreamed he was a butterfly, a butterfly flitting and fluttering about, happy with himself and doing as he pleased. He didn't know that he was Zhuangzi. Suddenly, he woke up. And there he was, solid and unmistakably Zhuangzi. But he didn't know if he was Zhuangzi who had dreamt he was a butterfly, or a butterfly dreaming he was Zhuangzi. Humans have long been questioning our senses in the search for a base reality. From Plato and the philosophers of ancient Greece, through India and the search for nirvana, all the way east to China and the yin and yang of existence, what we sense and what is truly real has been a topic of fascination for millennia. And in 1619, René Descartes joined that ancient struggle by sitting alone in a room and trying to destroy everything he knew. Descartes had decided that if any single belief in his head can be doubted, if there is even a hairline crack of uncertainty, it has to go. He wants to find bedrock, something so solid that no argument can ever shake it. And so he starts taking a hammer to the foundations of his own mind. The senses go first. He's dreamed vividly enough to know that what we see and feel and hear can be fabricated. A dream, of course, feels completely real while you're inside it, which means the evidence of his own eyes is worthless. Everything he thinks he knows about the world comes in through his senses, and his senses can lie. He goes further. What if there was a demon? He supposes an immensely powerful, immensely clever deceiver who had constructed a perfect illusion around him. A demon who feeds him every sensation, every memory, every feeling of solidity and warmth and the passage of time. All of it false, all of it designed to fool. And he realizes there would be no way to know. The deception would be perfect. Descartes couldn't disprove the demon. He couldn't disprove it because any evidence he might use to disprove it would itself be evidence the demon had provided. The trap seemingly had no exit. All he was left with, in the cold and the dark, after everything else has been demolished, is this. He was thinking, doubting. And doubting requires a doubter. The demon could fake a world. It couldn't fake the act of questioning itself. Cogito ergo sum. I think, therefore I am. And 300 years later, a philosopher named Hilary Putnam updated Descartes' demon with better hardware. Imagine, Putnam said, that your brain has been removed from your body and a supercomputer wired into every nerve ending feeds it a perfect continuous stream of signals. Every sensation, every memory, every conversation, every heartbreak, every ordinary Tuesday morning, indistinguishable from reality. Because as far as your brain is concerned, it is reality. How would you know? Of course, you wouldn't. You couldn't. Because every tool you could use to investigate the question, your eyes, your hands, your reasoning, your very sense of unease regarding the idea, would itself be part of the input. The brain in the vat has no outside view. And neither, of course, do you. And now here is where it gets even stranger. A cognitive scientist at UC Irvine named Donald Hoffman ran the numbers on this problem. However, not philosophically, but evolutionarily. He built computer simulations of organisms competing for survival. Some of his organisms perceived the world accurately. Others perceived only what was useful for survival, fitness signals stripped of any deeper truth. And the accurate ones lost comprehensively. The conclusion Hoffman drew is unsettling. Evolution doesn't select for truth. It selects for survival. Our senses were not designed to show us reality. They were designed to keep us alive long enough to reproduce. The apple looks red and round not because it's red and round at the level of physics. At the level of physics, it's a cloud of mostly empty space with vibrating quantum fields. It looks red and round because that simplified icon is useful, quick, actionable. In other words, the interface you're looking at right now, the world, is not the computer. It's the desktop. This is a more limited claim than Descartes' demon. Hoffman isn't saying that the world is fake. He's saying we've never actually seen it. What we perceive is a species specific user interface, tuned by millions of years of natural selection to hide the underlying machinery and show us only what we need to function. Reality is running in the background. We just don't have access to it. In a sense, he is saying that we are already living in a simulation of our own making. And so, if that is the case, then it's not a huge leap to say somebody could fool those inputs. And in 1999, the Wachowskis gave this idea a red pill and a black leather coat and made it into a blockbuster. The Matrix remains the most culturally potent version of the brain in a vat argument, the one that made millions of people, probably for the first time, genuinely wonder if reality could be a constructed experience. But here's what the film gets quietly wrong. When Neo is unplugged, when he wakes up in the grim machine world and gasps his first breath of real air, the film implies that he's now in base reality. But why would he assume that? Why would we? If a simulation can be perfect enough to fool you, then waking up inside yet another simulation would feel exactly like waking up. There's no pill for that. The skepticism has no floor. And this is where Bostrom's argument becomes genuinely personal. Not just we might be simulated in some abstract sense, but a specific, almost archaeological claim about which simulation we probably are. Bostrom called it the ancestor simulation. His reasoning was this. Advanced civilizations wouldn't just simulate random universes. They would simulate their own past, their own history, the evolutionary pressures that produced them, the civilizations that preceded them, the individuals whose choices shaped their world. For these advanced civilizations, these would be the most scientifically and philosophically interesting simulations to run, the ones that explained how they got here. And Bostrom speculates they would run them again and again and again. And this would mean that if simulated beings are possible at all, most of the conscious beings who have ever lived would be those simulated ancestors. Reconstructions run millions of times in millions of parallel branches by civilizations curious about their own origins. We would not just be living in a simulation. We would be the experiment, the historical reenactment, the digital zoo. And if that wasn't enough, there is an even more extreme version of this idea. Suppose we build a computer so powerful that it can simulate a universe as complex as our own. Within that simulation, those digital beings would also eventually become curious. They might build a simulation of their own. This is the grim concept of nested simulations. If it is possible to create one, then there were likely countless numbers of them stacked like Russian dolls. The natural universe, the one made of real meat and bone, would be buried under trillions of layers of digital replicas, each making their own digital replicas. Statistically, the odds that we are in the base reality, or anywhere near it, become vanishingly small. We would likely be just a subroutine within a subroutine. A dream within a dream being processed by a machine that we would never see. And so clearly, were we to inhabit a sufficiently complex simulation, it could fool our brains. We would have no way of telling the difference between an artificial world and the real one. No way of experiencing reality beyond our own tricked senses. But if that is what we are, if a first kiss quickens our pulse, if grief lands in our gut, if the sunset still takes our breath away, then in what sense is it any less? In a way, a perfect simulation is not a copy of the universe. It is a universe. What is reality at its most fundamental level if not information? This brings us to the ultimate challenge of physics, the question of what everything really is. And one man was on a mission to figure it out. The year was 1944. A worn postcard weighed heavily in John Wheeler's hands. It had been sent from Europe, from his brother Joe, on the front lines as Allied forces made their final push into Nazi Germany. The war was close to won, but Germany was still a potent and dangerous threat. Joe knew that his brother was working on something big. It was a secret project far beyond the pay grade of a lowly grunt. But Joe suspected, as did many other soldiers, that some of the smartest scientists in the world were gathered at laboratories and facilities scattered throughout America, all focused on a singular effort to produce a weapon so powerful that it could end the war entirely. And yet, every day, more lives were lost in the pursuit of ultimate victory. They were eager for any reprieve from the violence. And so Joe wrote only two words on that postcard. "Hurry up." And just a few months after Joe sent that postcard, he would be dead. John Wheeler was too late to save his brother. "Here we are so close to creating a nuclear weapon to end the war," he would later write. "I couldn't stop thinking then, and haven't stopped thinking since, that the war could have been over in October 1944." From that day, Wheeler would commit his entire intellect to solving the fundamental nature of reality. The past, Joe's death, was forever locked away from him, inaccessible. Why did the cosmos behave the way it did? What rules governed the universe, and how could we discover them? Indeed, Wheeler is perhaps the most influential physicist that is not really a household name. He was a postdoc under Niels Bohr and went on to mentor over 60 graduate students and over a 100 undergraduates and postdocs. Wheeler's core philosophy was grounded in radical conservatism. He based his research on known physical laws but pushed them to their utmost extreme. And so in 1976, Wheeler assembled a team of students and postdocs at the University of Texas to study the nature of quantum information. One of these postdocs was David Deutsch, who'd go on to kickstart quantum computing. But Wheeler was really interested in something even deeper. Where does reality come from? To explore this question, Wheeler took the familiar two slit experiment and showed just how absurd reality can be. If you send a line of photons through two parallel slits, you can't know which one it passed through. What you observe, even after the photons have crossed over, is an interference pattern on an opposite screen, exactly as if waves of water had passed through the device. But if you cover one of the slits, the wave pattern goes away and the photons act like a particle. A bizarre outcome of quantum principles, showcasing the wave particle duality of nature. However, Wheeler went one step further and created a truly bizarre thought experiment. What if you timed it so that you started with two slits, then after you knew the photon had passed through, you closed one of them off? Would the photon know that you had closed one slit behind it? And if so, what would emerge on the observation screen? His conclusion, borne out later by experiment, is hard to swallow. If you close one of the slits behind the photon, it acts as if the slit had been closed the whole time, and the photon collapses into a particle instead of interfering like a wave. Just how is this supposed to work? How could a photon reach back in time to see what you chose to do well after it had passed through the experiment? Why did the act of observation affect both the future and the past? Wheeler's solution was information. "It from bit" was the phrase he coined for this philosophy slash physics. He viewed the universe as a sort of self-excited circuit where the act of measurement gave rise to physical reality. In this view, the cosmos would be abuzz with quantum information sloshing back and forth in entangled, superposed states. And it would be the act of observation that would turn these probabilities into concrete existence. For Wheeler, that was it. That was reality. A series of observations. Indeed, without observation, there is no measurement. The time and place don't matter. What is essential is information, and when it is measured. When Paul Dirac challenged this idea during a lecture, he asked Wheeler if we create the moon every time we look at it. "Yes," was Wheeler's blunt reply. We would never be able to build a coherent theory of physics from this radical idea. But the kernel of truth that remained was that information was key to physics. Energy, momentum, quantum states, and everything else within physics can all be cast in terms of information. Information shuttling back and forth under certain rules. For Wheeler, that was the end of the line. Information was the central pillar of existence. But is there any evidence that supports this worldview? To the skeptics of the late 20th century, Wheeler's "it from bit" sounded less like physics and more like poetry. After all, a bit is an abstraction, a choice between yes and no, one and zero. How could something so ethereal, so devoid of substance, give rise to the cold, hard weight of a mountain, or the searing plasma of a star? Well, one answer lay in a discovery made in 1961 by a German American physicist named Rolf Landauer. Landauer worked for IBM, a company that at the time was obsessed with the efficiency of its room sized mainframes. Indeed, Landauer wasn't looking for the secrets of the cosmos. He was looking for the limits of the machine. He wanted to answer a simple question. Is there a minimum amount of energy required to process a single piece of information? Through his work, Landauer realized that when you erase a bit of information, whenever you take a one or a zero and reset it to nothing, you must, by the laws of thermodynamics, release a tiny puff of heat into the environment. It's fairly straightforward. You are reducing one or zero, two states, down to a single state. Nothing. Entropy is therefore reduced. And since the second law of thermodynamics states that local entropy can never decrease, entropy somewhere else must go up to balance out the total entropy, then that entropy takes the form of heat. And this isn't merely a flaw in our wiring or the result of inefficient copper. It is a fundamental tax levied by the universe itself. If you want to delete a thought from a digital brain, the universe demands a payment in energy. This became known as the Landauer principle and was the first concrete evidence that information is not just about the physical world. It is a physical thing in its own right. Changing information has a price. It has a temperature. And so logically it follows that if information has a physical cost, then every interaction in the universe, every photon striking an eye, every electron hopping in a circuit, every chemical bond forming in a cell, is a form of computation. And in 2002, Nobel laureate Anton Zeilinger took the idea one step further. In the quiet laboratories of Vienna, he took Wheeler's radical dream of an information processing universe and translated it into a hard experimental reality. By peering into the heart of the most elementary quantum systems, such as the spin of a single electron, Zeilinger demonstrated that these particles are the containers for exactly one bit of information. This means for over a century we have grappled with why our universe is quantized, existing in discrete, irreducible chunks of energy and matter. Zeilinger's work suggests that physics is quantized because information itself is quantized. Nature doesn't come in packets because of an arbitrary physical rule. It comes in packets because a single bit is the smallest possible unit of truth the universe can hold. And so perhaps Wheeler's crazy idea doesn't seem so crazy after all. If the universe is constantly processing information, and that information has a physical footprint, then perhaps the cosmos isn't just a collection of objects that happen to have data. Perhaps the cosmos is the data itself manifesting as objects. Perhaps at the very limit of reality, the it doesn't just emerge from the bit. Maybe the bit is the only thing that was ever truly there. The key argument in favor of simulation theory is that if the universe is made of bits, then the universe is by definition computable. And if it is computable, then someone somewhere might have already built a machine powerful enough to run the program. Indeed, in a sense, we've already done it, at least partially. And so, if computers are capable of recreating the universe, then what's stopping the universe from being a simulation? John Wheeler spent his whole life arguing that information is not a representation of reality. It is reality. The bit is the thing. It's all one and the same in the it from bit worldview. Indeed, it's clear that at the moment we can't rule out the possibility, and so we have to return to the grim apocalyptic reasoning of Nick Bostrom. However, there are other ways to get around Bostrom's argument. One is the argument that we know for sure we haven't created simulated consciousness in our own past. Nor have we met any aliens who have demonstrated the capability, or indeed any aliens at all. Bostrom's entire argument rests on the ability to do it in the first place. And until we someday do it in the future, or we meet aliens who already have, then we can never truly know for sure if it is possible. Indeed, some scientists like David Kipping have attempted to calculate the probability that we live in a simulation based on the extremely limited data that we have access to. Namely, that we ourselves can't create simulated conscious beings and we haven't met anyone else who can. Unfortunately, his calculations arrived at almost exactly 50/50 odds, which isn't hugely helpful. And so all that remains from Bostrom's argument, all that's left after our magnificent advances in simulating the universe, and what meager scraps of truth are laid bare after decades of Wheeler's searching, is a rather unsatisfying answer. We might live in a simulation and we might not. And we're a long way off from knowing. Either way, you've been watching the entire history of the universe. Don't forget to like and subscribe and leave us a comment to tell us what you think. Thanks for watching and we'll see you next time.