Rizwan Virk - APPENDIX to Sim
Frequently Asked Questions (FAQ) About the Simulation Hypothesis
As I mentioned in the foreword, in the five or so years since I published the first edition of The Simulation Hypothesis I’ve interacted with thousands of readers in person, online via social media, and via email. While each discussion was unique, I found that certain questions came up again and again. Here is a list of frequently asked questions related to the simulation hypothesis, and my most recent thoughts on the answers.
DO YOU REALLY THINK WE LIVE IN A SIMULATION?
Yes. I wouldn’t have written this book if I didn’t think we were in some kind of simulation. However, there are subtleties.
The real difficulty in answering this question (and the next one) is that it depends on the type of simulation we are living in. I often say the simulation hypothesis is several different ideas wrapped into one—some of these ideas aren’t that controversial, while others are highly speculative, and some seem positively outlandish.
For example, one of the key ideas in this book (you might even have gotten tired of me reiterating this point again and again) is that the universe is made of information rather than matter, and this information is computed and then rendered for us into what we think of as the physical universe. I recently met a Nobel Prize–winning physicist at the University of Cambridge in the UK and he suggested that the first part of this statement, that the universe consists of information, is no longer controversial.
Most people don’t realize how far this idea has come in the world of physics—there is a whole new branch called digital physics, which I gave a definition of in Chapter 14. When I was a kid, we were taught laws like the conservation of energy, the conservation of momentum, and the second law of thermodynamics. These were classical physics ideas, and though quantum mechanics had been around for at least fifty years when I was in school, most people still thought of the world as classical (and still do today, if we’re being honest).
Yet today, we have the principle of “conservation of information,” as opposed to “conservation of momentum” or “conservation of energy,” as physicists start asking questions about how much information is in the world, whether it can be destroyed, and what happens if information is lost.
The universe seems to be some type of computation machine. As I went into some detail in The Simulated Multiverse (and mentioned briefly in Part IV), physicist turned computer scientist Stephen Wolfram is working on a method that uses cellular automata as a model for how the universe works, based on computation rather than mathematics. And in Chapter 14, I also described Vopson’s proposed second law of infodynamics, about what happens to information entropy versus physical entropy as a system evolves (which also seems on the surface to correspond with Campbell’s fundamental process, at least in the profitability function for biological systems being a state of least entropy, described in Chapter 8). In the past, physics was described using mathematics, but in the future, it may be described by computation.
Another part of this statement is that the universe consists of information rather than matter; this part may be a little more controversial, but perhaps not as much as it seems at first blush. I often make the point that the more you look for this thing called “physical matter,” the harder it becomes to find, particularly if you ask physicists. This desk that my computer is sitting on right now while I type this is made of more than 99 percent (perhaps even 99.99 percent) empty space. Then when you look into the molecules and the atoms, you also find…a large amount of empty space. Like with those nested Russian dolls (called matryoshka or babushka dolls), you keep looking at smaller and smaller definitions of matter, and what do you find at the bottom level? According to physics you find a particle. But what is a particle? This seems like a simple question that we all know the answer to, but it turns out we don’t have an easy answer. As I mentioned in Chapter 15, John Wheeler came up with the phrase “it from bit” because he came to the conclusion that at the “bottom” level, the only thing that distinguishes one particle from another is answers to yes/no questions, or a series of bits.
So, once again, the idea that the universe consists of information rather than matter isn’t particularly controversial, even among some physicists. The question of how that information gets presented to us as the “physical world”—that is, space and time—is still not really understood, except that the brain accepts sensory input and then processes it into the world as we see it. To me, the metaphor of a video game, in which you take information from the server and then present it, seems like the best explanation for how this might work. And if we are all doing the same thing, taking in information and processing it, augmenting it, and most importantly, rendering the physical world, it suddenly becomes like a massively multiplayer online role-playing game.
This perspective is more controversial; some think there is still a physical world but that our perceptions of it may not be correct, so we get an inaccurate view of this physical world through our sensory inputs. This puts us right back into the world of philosophy (Descartes’s evil demon, for example, or Donald Hoffman’s original metaphor of the world as a desktop, described in Chapter 14).
Whether this is a computer-based video game, in the way that we understand computers to date (i.e., classical computing using the von Neumann architecture), is an open question, and is where some of the debate comes in. Two prominent researchers in quantum computing, MIT’s Seth Lloyd and Oxford’s David Deutsch, have made the point that the universe is computing all the time. Deutsch modified Wheeler’s phrase to “it from qubit.” Both quantum computing and qubits, described briefly in this book, are described more fully in my book The Simulated Multiverse.
The second big idea that is part of the simulation hypothesis, which both derives from and informs the first big idea, is that the world around us isn’t real—this goes further than saying it’s just a user interface, insisting that the universe as we see it is a kind of purposeful hoax, an illusion. This idea comes to us through both philosophy and religion. In fact, among the world’s religions, as I’ve shown in Part III, this idea isn’t controversial at all. I like to call it one of the most fundamental articles of faith among the world’s different religions.
If we take these ideas together, that the physical universe consists of information that is rendered for us, and that the world is some kind of purposeful hoax, we get the essence of what I like to call the simulation hypothesis. Either flavor, the NPC or the RPG version, and the variations in between, all fit into this model. If you honestly research both of these big ideas, I think you will come to a similar conclusion to mine: The physical universe as we know it is some kind of simulation or video game, a counterfeit world that is built on information and is not exactly what it seems.
WHAT ARE THE ODDS THAT WE LIVE IN A SIMULATION?
I have made the statement in the past and stand by it that we are at least 51 percent likely to be living in a simulation—that is, we are more likely to be in a simulation than not. When I made this statement in 2019, shortly after the release of the first edition of this book, the story went viral, saying that an “MIT professor” claimed we were living in a simulation.
Before and after that, there have been many attempts to put a percentage on the likelihood that we are in a simulation. Each one has used different methodologies, though each has relied on Bostrom’s simulation argument to a lesser or greater extent. In the five years since my original pronouncement, I have updated my estimated percentage likelihood that we are in a simulation. Before I give you that updated percentage, I’ll give a quick overview of the other attempts.
In Chapter 6, I outlined Bostrom’s simulation argument and his three possibilities whose probabilities must add up to 100 percent (i.e., one of the three possibilities must be true). I like to call this Bostrom’s trilemma, and I am restating them here, but rather than using his original terminology (which included posthuman), for reasons I outlined earlier, I am using my own terminology of the simulation point. This is because my terminology includes not only the ability to have simulations with posthuman AI but also the ability to include avatars of “real” characters. Rather than stating them as probabilities that must add up to 100 percent, I will present the likelihood that we are in a simulation if that leg of the trilemma is true:
Option 1: No civilization ever reaches the simulation point. Chances we are in simulation if true: zero percent (0%).
Option 2: All civilizations that reach the simulation point decide to ban the creation of “ancestor simulations” and don’t create any. Chances we are in a simulation if true: low, almost zero ( ~ < 1%).
Option 3: One or more civilizations reach the simulation point, and each one creates a large number of simulated universes, which are indistinguishable from our own.
Chances that we are in a simulation if true: We are most certainly in a simulation (~99+%).
You’ll notice that Bostrom himself, in his original paper, didn’t put any percentages on either of these three possibilities. It is usually assumed that that option three is Bostrom’s whole argument—that he concluded we are most certainly in a simulation (to be fair, he did use the nomenclature “the simulation hypothesis” only when referring to option three in his trilemma). When asked to put a percentage on it, Bostrom initially stated that he would consider the three possibilities equally likely—that is, 33.333 percent each. Later, when pressed to provide an updated answer, Bostrom said that he would put option three at slightly lower than the other two, in the range of 20 percent, but this was based mostly on intuition.
This is of course at odds with folks like Elon Musk, who said in 2016 that the chance that we are in base reality (i.e., that we are not in a simulation) is one in billions. That would mean that the chances we are in a simulation are billions to one (or technically “billions minus one,” but that still isn’t a precise number); let’s just say he puts it at 99.999-plus percent. However, technically, Musk’s argument relies on option three from Bostrom’s simulation argument, so we would need to multiply 99.999 by the percentage likelihood of option #3 (33.333 percent, using Bostrom’s first estimate), which is still pretty close to one-third, depending on how far you take the nines and the threes in decimal points.
In 2020, Columbia University astronomer David Kipping published a paper that tried to estimate the chances that we are in a simulation using Bayesian reasoning, and he put it at slightly less than 50 percent, a number that corresponded with astronomer Neil deGrasse Tyson’s estimate of 50 percent.
David Chalmers, who wrote his original paper about the Matrix Hypothesis in the same year Bostrom published his paper (2003), says that if the simulation were perfect, there would be no way to know if we were in a simulation or not. In my mind, this is the same as saying it’s a coin flip— or that we are 50 percent likely to be in a simulation (though Chalmers didn’t explicitly state this percentage).
You will see in each of these formulations that they require someone to build the simulation (see the next question, “Who created the simulation?”). For this to happen, they’d have to get over the hurdles or gates that exist in reaching the simulation point. In other words, it must be possible to reach the simulation point in order for the percentage to be more than zero.
My own updated percentage is based partly on these ideas, but the last point is the key. If it is possible to build a Matrix-like simulation—that is, if it is possible to reach the simulation point—then it’s likely that someone has already gotten there and we are already inside a simulation.
If video games, their interfaces, and the AI in those games all continue to improve, then we will reach the simulation point for sure. If we are 100 percent likely to reach the simulation point, then we end up pretty much in Bostrom’s third proposition: We are most certainly in a simulation (ignoring for the moment the branch that bans simulations). Or, as Musk said, the odds are infinitesimally small that we are in base reality.
So, basically you can see the whole argument goes from 0 percent to almost 100 percent depending on the likelihood of reaching the simulation point. To date, we have only one technological civilization to look at: ours.
So how close are we, the only technological civilization we have as data, to the simulation point? I would estimate that we are at 67 percent, or two-thirds down the road to the simulation point.
There are ten stages on this road to the simulation point (see Part I); however, this doesn’t mean that we are 100 percent of the way through seven stages and 0 percent of the way through the remaining three. Rather, it is an average amount of how far we are on each stage (see the question “When will we reach the simulation point?” for more details).
Thus I would say that we are, as I write this in 2024, in the neighborhood of 70 percent likely to be inside a computer simulation. This is, of course, based primarily on the technological argument, but it ignores the other clues that we are in a simulation—such as the quantum mechanics and physics weirdness, and the reports of all the mystics.
Thus, my revised answer to this question in this second edition is at least 50 percent, and most probably closer to 70 percent likely that we are in a simulation.
WHO CREATED THE SIMULATION?
This one is in the list of top questions that I get asked. Another way to ask it is: Who are the simulators?
I tried to answer this in Chapter 15. However, if we look at Bostrom’s original argument, he was mostly concerned with future humans who might create ancestor simulations. This would be akin to us creating a simulation of ancient Rome or Greece, for example; in a sense, we in the West are descendants of the Romans (or the ancient Hindus, or others), so they are our ancestors.
However, calling them future humans can be confusing, because it implies they don’t exist yet. In the case of an ancestor simulation, we would be simulations of the ancestors of the people making the simulation. Theoretically the simulation will end when it catches up to the people who are building it, though it could go on to simulate the future if the simulators mess with time inside the simulation to make it go faster (a distinct possibility).
It’s for this reason that when discussing an ancestor simulation, rather than calling the simulators “future humans,” I like to simply call them a “more advanced species.” This could be a species of human (on an Earth outside the simulation) or it could be some nonhuman intelligence (aliens?) who decided to simulate our civilization. Note that this is true for both NPC and RPG versions. This also means that an accurate ancestor simulation becomes a form of simulated time travel, which you might feel if you go into a game like Assassin’s Creed.
While the majority of avatars we create in simulations are human, there are a considerable percentage of characters that could be another species altogether. In fact, the purpose of the sim might be to see how humans as a computational species might evolve and what they might accomplish, collectively or individually.
In The Matrix, the creators of the simulation are actually superintelligent AI. That is another popular theory—that the creators of the simulation are not humans at all but rather the machines we left behind. Why would they simulate us? We can speculate on this, though the doomsday scenario depicted in The Matrix—that they are using us as batteries—seems unlikely and unnecessary to me.
I take the idea that we are the creators more seriously than most academics, but we may not be biological (human or alien) at all. We might be spiritual beings without physical bodies looking to have a corporeal existence, and this is why I dedicated Part III of this book to looking into parallels between the simulation hypothesis and the world’s religions.
This holds open the idea that we are the creators or at least architects of our fate, of what games we are going to play. There may also be more advanced players, and in a technological twist on traditional theology, there may in fact be a single creator (or one we would consider God) or creators. Because simulations generally show signs of intelligent design, this seems to me to be a reasonable conclusion and once again shows the relationship between simulation theory and the world’s religions.
WHAT IS THE PURPOSE OF THE SIMULATION?
I often get asked this question, and I like to answer it by asking two related questions:
Why do we play video games?
Why do we run computer simulations?
Though these two questions may seem different at first glance, I think they hit upon similar underlying themes, which have to do with the costs and benefits of running simulations. Let’s take them one by one.
The reason we play video games is usually some variation of “because they are fun.” The reasons they are fun, though, can vary depending on who you ask and what types of games. As for the more realistic games, like 3D MMORPGs, my answer to this question is that they allow us to have experiences (and make choices) that we can’t have in the physical world. I always like to joke that I can’t fly on a dragon and shoot arrows at orcs in the “real” world, but these are things my character could do in a video game. Similarly, most of us can’t (and wouldn’t if we could) do illegal things like stealing cars and hurting others in games like Grand Theft Auto. We also can’t explore other planets (not yet!) or fly spaceships and have laser battles in outer space, but we can do these in space-based MMORPG games. In a sense, these kinds of games, whether they are single-player fantasy RPGs or multiplayer virtual worlds, let us have experiences that we can’t have in the “ordinary” world outside the game.
This includes collaborative group and team experiences, and one-onone relationships with other people. When I used to spend time inside Second Life, one of the first successful virtual worlds or metaverses, I found that players logged in to have interesting social experiences and a host of virtual relationships. In some cases, this included having a virtual job in a nightclub as a bartender, for example. They got paid in virtual currency, but it was peanuts. It was clear that it wasn’t just for the money, but because these virtual jobs gave them a sense of purpose in the game and perhaps gave them a level of socializing they may not have had in their normal, dayto-day jobs. Unlike traditional shooter or action games, which tended to skew male in gender demographics, virtual worlds were almost a fifty-fifty split between male and female, emphasizing that they are actually virtual social worlds.
I often acted like a virtual anthropologist, asking the players or characters I met what they were doing in real life, what they liked to do in their virtual lives, and why. In some cases, women (and men) who were married decided to get married again online (to someone in another part of the world), while others who were married IRL (“in real life”) decided to be promiscuous and explore many partners in the virtual world. Some visited Star Trek or other video-game-themed sims, while others went about as furries (avatars that are based on animals like cats, dogs, etc., but with anthropomorphic forms). Some liked to role-play certain time periods or fictional worlds. In many cases, they wanted to have experiences that they might not have been able to have in the real world; in others, they wanted similar experiences but with different people from their normal, everyday lives. In a few cases, they ended up actually meeting their virtual romantic partners IRL, with mixed success.
In the end, the idea that video games (and virtual worlds) can give us experiences that we don’t have IRL or outside the game leads us to at least a partial explanation for the purpose of the RPG games: to have experiences. As I said in Chapter 12, the world is meant to be a kind of game, according to various religious traditions. In Arabic, the term the Koran uses for the game is al ghouriri matau—an “enjoyable delusion”— and in the Hindu traditions they use maya to designate a “carefully crafted illusion,” but also lila, the divine play of the gods. Perhaps the experiences we can have here, of pleasure and pain, of the relationships and emotions we experience in the game, are things we cannot experience in a spiritual world.
This includes the unpleasant experiences—whether they are social, like betrayal, breakups, anger, or challenges like physical ailments and financial troubles, not to mention wars and suffering in general. I will say that this provides a framework or way to think about how our experiences in the physical world might be valuable, simply to enjoy even while we experience a gut-wrenching gameplay session. I always like to make the point that when actors are nominated and win an award like the Oscar, it’s often when they are playing roles that have gone through some serious challenges and a difficult but rewarding life.
The second question is about the purpose of running simulations. Whenever we run a simulation, it is to figure out what might happen—in other words, what is the most likely outcome, or perhaps the most (or least) desirable outcome. The reason we run computer simulations is so that we can run not just one but multiple versions—usually we end up running a simulation once, and then we change the variables and run the same simulation again and again before we decide which outcomes are most likely or most desirable.
Simulations are useful for processes in which there can be many independent variables and it’s difficult (or impossible) to know how a process will turn out after so much time has passed (a year, five years, ten years, etc.). Unlike stable or periodic processes, in which you can predict what will happen at step 2,000 based on what is happening now, for chaotic and complex processes it is hard to predict. The only way to know what will happen at step 2,000 is to change initial variables and go to step 1,999. But the only way to know what step 1,999 might look like is to go back and simulate up to 1,998, and so forth. In other words, these processes are computationally irreducible—you can’t reduce them to a simple formula. One of the most famous examples of this type of process is the three-body problem—in which three bodies are gravitationally in a system, like a triple-star system. Will it reach a stable configuration? Will one star smash into another? Will one go flying off into the galaxy? It’s hard to say without simulating what will happen.
Coincidentally, when I interviewed legendary science fiction writer Philip K. Dick’s wife, Tessa, she brought up his speech from Metz, France, in 1977. She encouraged me to look beyond the phrase that’s become popularized, which I quoted earlier (“We are living in a computerprogrammed reality…”), and to look at the rest of that sentence, “and the only clue we have to it is when some variable is changed, and some alteration in our reality occurs.” The speech, titled “If You Think This World Is Bad, You Should See Some of the Others,” was a fascinating artifact, in which Dick tried to articulate his theories about not just being in a simulation, but beings that were able to change the variables of our simulation and rerun timelines. Each timeline was like another world. Tessa said Dick came to believe that his book The Man in the High Castle, in which the axis powers of Germany and Japan won World War II, was a real timeline that had been “unwound.” I wrote The Simulated Multiverse as a thought experiment kicked off partly by thinking about Dick’s speech and one of the core ideas therein: that we could change variables and run multiple timelines in a simulation.
The key point here is that the purpose of the simulation, in many cases, is to see what the outcome of a computationally irreducible process is—and the only way to do that is to run the simulation. Whether we look at the RPG version or the NPC version, answering these two questions can give us a good first answer to the question of why we might be in a simulation.
WHEN WILL WE REACH THE SIMULATION POINT?
When I wrote the first edition of this book, I thought the simulation point was, at best, a few decades away, and at worst a century (or two) away. Even if it was twice this time, or even a thousand or ten thousand years in the future, in cosmological time this is still only the blink of an eye. As long as we don’t destroy ourselves in the process, we are likely to reach the simulation point in the future.
In fact, if we look at normal science fiction assumptions, or even normal assumptions made about our galaxy (assuming for a moment that we are in a physical universe), there would likely be civilizations that are a thousand or ten thousand years more advanced than us. For example, in well-known sci-fi writer Ursula K. Le Guin’s Hainish Cycle series, there are the remnants of an advanced civilization that has been around for a million years.
We can conclude it’s reasonable to assume that in a physical galaxy, many civilizations will have already reached the simulation point and Bostrom’s conclusion that we are most likely in a simulation is valid.
As I said, my best estimate is fifty to a hundred years, but at the pace at which we are building AGI, many experts believe it will be here much sooner, possibly in a few years. Given how fast video game and AI-based rendering is also expanding, we can assume that as soon as we hit humanlevel conversational ability with chatbots (which many think we already have as I write this in mid-2024), they can be paired with humanlike avatars and the AI system will feel real, passing the Metaverse Turing Test, like the virtual girlfriends in Her and Blade Runner 2049. In 2024, we have many people turning to AI companions as virtual boyfriends and girlfriends, as we discussed in Chapter 5.
In any case, I believe that while AI will be able to easily do many tasks that humans do within this decade, true humanlike AI, which can learn from its environment (in a virtual world) and have self-awareness of its virtual body without major flaws, will take a few more years to complete. I believe we are in the second great wave of AI; the first was rules-driven expert systems (and back in the 1980s they were predicting we’d get something like AGI by the end of that decade). This second wave has been driven by data (rather than rules) and by supervised and unsupervised training of models like LLMs (large language models), which make it possible for a chatbot like ChatGPT or Google’s Gemini to put forth realistic text generation. In other cases, today’s AI uses diffusion techniques to generate new images from its training set data. LLMs don’t technically follow rules and can’t calculate.
That said, I believe a third great wave of AI is arriving soon, which will combine machine learning with rules-based reasoning, an AI architecture analogous to a von Neumann architecture (which is the standard architecture used within computers to date). In fact, in September 2024, OpenAI previewed their newest model, which they claim can already do humanlike reasoning. The result will be agents that not only do what they were trained to do but also make judgment calls based on logic and can implement existing and new rules. As this book goes to press, the CEOs of OpenAI and Anthropic have gone on the record stating that AGI will be available as soon as 2025 or by 2027. As you are reading this book, you can be the judge of how close we have gotten.
I believe we are at least five years and possibly a decade away from true AGI virtual characters that can pass the Metaverse Turing Test. I have already seen some demos that look pretty impressive. We also still have to get over the hurdles to understanding how best to build brain–computer interfaces (stages 7 through 8) to get to the simulation point.
I think BCIs to read and write may be among the most difficult stages, so they will take us at least another decade. It’s not clear how long downloadable consciousness will take (or even if it can be done, but let’s assume it can for the purpose of this discussion). Therefore, I think that to get to the simulation point will easily take us to 2049 at least and possibly up to 2075, which would be at most fifty years from now. In any case, even by the most conservative estimate, I believe by the end of the twenty-first century we will have reached the simulation point.
IS LIFE MEANINGLESS IF WE LIVE IN A SIMULATION?
Some people view the simulation hypothesis as a bit nihilistic, or even solipsistic. They say that believing we are in a simulation would justify selfserving or incredibly bad behavior. If you’re not sure what those terms mean, don’t worry, I had to look them up, too, in order to make sure I’m getting them right.
nihilism: “rejecting all religious and moral principles in the belief that life is meaningless”
solipsism: “the quality of being very self-centered or selfish,” or in philosophy, “the view or theory that the self is all that can be known to exist”
First, I would make the point that you don’t need simulation theory for people to justify incredibly self-centered or bad behavior—you just have to look around at the world today. Whether it’s individuals, companies, or nation-states, self-centered behavior seems more the norm than not.
I would, however, emphasize that both of these views point at an important distinction between different flavors of the simulation hypothesis: the NPC versus RPG axis. As I’ve mentioned, this is an axis that goes from 0 to 100 percent, and each of these flavors is sitting at one end of the axis. If we consider the number of players in the game:
n = 0. In this extreme case there are no live players (i.e., entities who exist outside the simulation and have avatars in the simulation). This means the simulation would consist of 100 percent NPCs.
n = 1. In this case, there is one player in the simulation and everything that happens is for the benefit of this one person.
n = 2 to 99%. In this case, there are multiple players and many NPCs, depending on where in the range (low end to high end) you want to put the pointer.
n = 100%. In this extreme case on the end of the spectrum, everyone is a source player and there are no NPCs.
You’ll notice that a nihilistic response is akin to thinking there are zero source players and that everyone (including the subject in question) is an NPC and just computer code. If no one is “real,” then this leads to the idea that nothing matters—and if nothing matters, you can do anything bad in the simulation.
The solipsistic view is that I am the only real person in the simulation, which is akin to the one-person simulation, or as we call it in the industry, single-player game. In fact, many video games that were played on your computer or console were originally single-person (or two-person if your friend is sitting next to you). This also leads to the conclusion that you can, just as you might in Grand Theft Auto, mistreat anyone because they are likely NPCs.
The height of this thinking is known as the “Matrix defense,” which was considered as a possible defense by Joshua Cooke, a nineteen-year-old who killed his parents and blamed it on watching The Matrix too many times—if the world is a computer simulation, then it shouldn’t matter. Legal experts have classified this as a kind of insanity defense. It’s important to note that the Matrix defense didn’t work in court, for Cooke or anyone. Astute readers will see that argument does rely on more of an NPC flavor of the simulation.
However, as you get toward the RPG side of the axis, toward a view that all the people you meet (or at least most of them) are real source players, then how you treat them matters. In fact, the RPG version, as I’ve pointed out in Part III of this book and in numerous presentations, is not that different from most of the world’s religions, which tell us that we exist outside the physical world as souls and that we have avatars, or characters, inside the video game. Despite this structure, killing is more or less the worst thing you can do on Earth according to any of the religions. It is one of the Ten Commandments that God gave to Moses, and pretty much occupies a unique place among sins or bad karma in other religions.
Even within Hinduism and Buddhism, which contend that you will simply “wake up” and go into another life if you die, killing someone is considered extremely bad karma and will affect you for many lives to come. We have the well-known Buddhist story of the goat and the priest (I first saw it at the beginning of the movie Little Buddha, in the form of a children’s story being told by a Tibetan lama). In the story a (non-Buddhist) priest is about to sacrifice a goat and notices that the goat is smiling. When the priest asks the goat why he is smiling, the goat replies that he was a goat for five hundred lifetimes and is now finally going to “level up” (my language, not in the ancient story!) and have a life as a human. Then the goat starts to cry. When the priest asks why the goat is crying, it replies, “Because like you, I used to be a priest who would kill goats, and so I had to live five hundred lives as a goat.”
The main point I would like to make is that if you think living in an RPG simulation makes it OK to do bad things to people, then you are essentially saying that the world’s religions make it OK to do bad things to people. In fact, if you thought that, you would have missed the main point of all the world’s religions and the main point of the RPG version of the simulation hypothesis! Of course, here I’m referring to what the religions say, and not what adherents of the religions actually do—they are just as likely to be self-centered or tribe-centered and commit atrocities as anyone else.
As I said in Part III, the life review is one of the key components of the near-death experience that added credence to the idea that we live inside a 3D world that is being recorded and generated on demand, and from which any scene can be replayed. More important, this review of the “game” of life pretty much gives us the purpose of life: Because you have to relive each moment of your life but from the point of view of other people, the real purpose of life is to be kind to other people (and other living things) and treat them well. In fact, NDE-ers who’ve had life reviews tell us how the little moments—smiling at someone when they are having a bad day, giving someone a compliment, watering a plant—often matter way more than what we think are big things like giving to charity. And treating others badly, emotionally or physically, is an experience that you will have to relive—from their perspective.
Moreover, the NPC model is, in a sense, comparable to the materialist view of reality—there is no afterlife, you know only what you have learned in the material world, and when you are gone, nothing you did will matter. And yet, many scientists who subscribe to a materialist, atheist point of view are among the most moral people I know. Which shows that you don’t have to be religious to be moral or to treat people well. In my opinion, the simulation hypothesis is not a license to not treat people well—in fact, if you believe in the RPG version, then it is exactly the opposite.
There is also a perspective that just because something happens inside a game or a virtual reality does not mean it doesn’t matter or it isn’t “real.” Suppose you were to learn a language in a virtual reality. Would that not matter to your life either in the virtual world or outside it? David Chalmers, the well-known philosopher who has written about virtual worlds, made the point in his book Reality+ that just because it’s virtual doesn’t mean it’s not real. You can have a real conversation with people in the virtual world, and the things you say and do matter.
Moreover, believing that the physical world is a game can be comforting in a number of difficult situations, whether physical, financial, or social. You can believe, like the actors who win Academy Awards, that you have chosen some difficult roles, or difficult quests, and that you have learned what you need to learn from these experiences and are ready to move on.
IS IT SIMULATIONS ALL THE WAY DOWN?
One question I often get asked is about stacked or nested simulations. This is when one simulation creates another simulation inside it. You’ll notice that if we are in fact inside a simulation (and this book has been about all the reasons why this may be true), and if we reach the simulation point, then we will, as a technological civilization, be able to create subsimulations, or child simulations, or sims within sims.
This is a complex subject, one that has also been explored within science fiction, and once again it depends on where you fall on our simulation being on the NPC-versus-RPG axis.
One way this subject is brought up is to try to disprove Bostrom’s simulation argument. You’ll recall in the simplified version of his argument, a technological civilization will create many simulations (in Bostrom’s original paper these are likely to be ancestor simulations, but there is no reason why they have to be). The simplified argument says that there are many simulated worlds and only one base reality (see Chapter 6).
Now, in the original formulation, and at first glance, this seems to be a case of having multiple parallel simulations. But if sims can create sims within them, or stacked simulations, then each of these will create child simulations, and each of those will create child (or grandchild) simulations, leading to a treelike structure. The basic argument still holds, it’s just that rather than having parallel simulations you have nested simulations.
This does lead to several complications. For one thing, if you think of computing power in a classical computing environment, each of the child simulations will have less computational power and, theoretically, less complexity than the parent simulation. It’s not clear if this would be the case with quantum computing, because the idea would be to use the same computing device as the parent universe.
The argument put forth by some, including physicist Sean Carroll, against being in a simulation is that in such a tree, most simulations will be on the bottom level, and these would be extremely low-fidelity simulations. This means that we are most likely to be not just in a simulated world but also a low-fidelity simulated word—that is, we should theoretically be in an 8-bit world that looks a lot like the old 8-bit games (think Pac-Man or Space Invaders). Therefore, because the world around us is complex, we are probably not in a simulation.
Or, in a variant, creating infinite child simulations will use up the computing power of the computing system in base reality, resulting in what programmers often call a “stack overflow error.” This was a common error when we all used to write our programs manually and wrote a recursive algorithm that went on infinitely; the computer ran out of memory. Note that this doesn’t happen simply because you run the same operation an infinite number of times. It happens if each time you run it, you use up more memory, and that memory is not cleared (what we call garbage collection), so that eventually there is no memory left. That’s why it’s called a memory leak.
This infinite stack overflow error could be a good reason why infinite simulations aren’t allowed. Unlike physicists, who often rely on infinity in their arguments, even Bostrom, in his original argument, relies not on infinity, but on a finite (but large) number of simulations. For one thing, there is no reason child simulations could always be allowed to create grandchild simulations. The process doesn’t have to be infinite, and there could be limits as to what is allowed. In fact, this may even be a clue that we are in a simulation, if computing resources were to dwindle in lower simulations and we were not able to create child simulations.
If we reach the simulation point, we can create simulations that are indistinguishable from our physical reality. That doesn’t mean they are as complex as our physical reality, because of the optimization techniques that can be used. Today’s high-fidelity video games and AI-generated videos are getting pretty close to photorealistic and appear to be just like our world in many cases, but that doesn’t mean they have the exact same complexity.
Moreover, just because our universe seems incredibly complex doesn’t at all mean it is at the same level of complexity as base reality. To use a simple example, when games and computers were 8-bit, the colors available were limited to a small palette. In fact, with 8 bits you can basically have 256 different values. In the 1980s I remember when more sophisticated computers came out and they advertised the ability to have millions of colors. This seemed like an impressive number. Of course, the ability to create lifelike images is not just about the number of colors, but also the number of pixels and the ability to simulate depth.
If you were inside an 8-bit game, you might not realize that you weren’t seeing all the colors that were available unless you stepped outside the game. Has this happened? Again, I turn to those who have had NDEs (see Chapter 12). Many NDE-ers report that after leaving their bodies they went to a world (which some call heaven but let’s call it, as many religious scholars do, ultimate reality, which in simulation speak we might call base reality) that has significantly more colors and sounds than are available in our physical world. In fact, when some of them return they report how “that world” was “more real” than this world—it was more vivid and colorful, and it had shades of subtlety and frequency in the form of music that we can’t reproduce in our limited “fake” world. This does in fact imply that base reality might be more complicated and have higher resolution and fidelity than this reality. It does not necessarily mean there are multiple nested levels beyond that (though there certainly may be!). We could be in a set of stacked simulations, but they may only be able to run several levels deep.
Finally, the ability to create nested simulations may be a key part of any simulated environment. In fact, going back to the question about the purpose of the simulation, it could be to see how many, if any, of these nested simulations are able to reach some key filter or singularity—to get off the planet, to blow themselves up, to reach artificial intelligence, or perhaps even to reach the simulation point.
I will bring up a science fiction example to see how or why it may or may not be allowed to create nested simulations. In the film The Thirteenth Floor, which may in fact be a better representation of ancestor simulations than even the more popular film The Matrix, we learn at the end that the simulation of 1999 was only one of thousands. However, it was the only simulation out of those thousands that was able to create its own nested simulations, and the simulators (who were in the year 2024, which incidentally is the year I am writing this second edition) were going to shut down the simulation because it was using too much of the computing power.
This ability to affect the computational environment of the parent world (whether it’s a simulation or not) may be an important clue, and it may in fact be what we do with our nested simulations. This also brings up an important twist on the simulation argument. Normally, I would say that if we can reach the simulation point, then it is more likely than not that we are in a simulation.
However, what if we can’t reach the simulation point? Using the same logic as before, one would assume this means that we aren’t in a simulation. Yet, the rules around stacked simulations might belie this assumption. Even if we cannot reach the simulation point, this might be an indication that we are in a simulation and either the simulators have limited our ability to reach this point, or we don’t have the computing power remaining to create simulations that look like ours. Both of these reasons imply that even if we can’t reach the simulation point, we might still be in a simulation!