No, You Don’t Live Inside a Computer
Why Elon Musk is Wrong
He’s now officially the world’s (second) richest man, but that doesn’t mean he knows everything. Elon Musk has infamously espoused that we live in a computer simulation. Indeed, he claims to believe that each of us is nothing more than a few bits of data, jumping around on the hard drives of some superhuman race. But despite his renown as an Earth-Saving, Tunnel-Digging, Space-Mining Supergenius, Musk seems to fall a bit short of expectations when it comes to metaphysics… Here’s why he’s is wrong.
The Simulation Hypothesis
Elon’s fantastical claims of computer simulations are a version of what I’ll call the simulation hypothesis. Generally, this position asserts that our entire universe exists inside a computer that is run by some incredibly advanced civilization. While support for the simulation hypothesis comes in many flavors, here’s the general form (n.b. I will describe a popular version of the simulation hypothesis, one that Musk seems to defend. See here for the original):
At some point in the future, human beings will develop powerful enough computers to run simulations of the entire world. That is, future humans will be able to create virtual universes inside their computers, much like video games. To begin, these virtual universes will be equivalent to what the real universe looked like at the beginning of time (or, for the physics purists, at some major event such as a big bang). Then, the future humans will press play on their computers and watch as everything that has ever happened unfolds on their screens. They’ll witness a simulated version of the solar system being born, they’ll marvel at the first simulated lifeforms, they’ll laugh as simulated cavemen clumsily grapple with fire, and they’ll cry as simulated atrocities occur throughout simulated human history. As time passes, the human beings inside the simulated world will become more and more similar to the future humans themselves (i.e. those who are running the simulation). Indeed, after long enough, the simulation will catch up to reality: the simulated humans will inevitably become the future humans and, just like their predecessors, will want to run simulations of their own. Thus, the simulated humans will build simulated computers, run a simulation inside their simulation, and watch as doubly simulated humans begin to run simulations of their own. And so the process continues indefinitely: as each group of simulated humans evolves and runs their own simulations, they will create deeper and deeper simulations, each nested one inside the other, ad infinitum. Therefore, the simulation theorists claim, there will one day exist a nearly infinite number of nested simulated worlds, with only one “real world” at the top of everything (we’ll call this real-world “base reality”). Further, because all of the worlds (both real and simulated) are so similar, there is no way for us to tell which one we’re in! But, simulation theorists say, because there are potentially trillions and trillions of simulated worlds and only one base reality, it is overwhelmingly likely that ours is one of the simulated worlds. Therefore, it is only rational to believe that we live in a computer simulation.
A bit of a problem…
The simulation hypothesis often inspires a sense of lonely, futile dejection. After all, who wants to live in a simulation?! Base reality seems far more desirable. But fear not. While the simulation hypothesis has so far convinced many worthy minds of its soundness, I will offer a principled reason for believing that it is not true.
First, we must grapple with what it would actually mean to run a simulation of the entire universe. In the old days, physicists and chemists thought that everything in the world was made up of tiny, indivisible pieces of matter called atoms. While scientists today believe that there do exist particles that are smaller than atoms, the notion that there nonetheless exists some smallest unit of matter that itself cannot be further broken up is still generally accepted. What exactly constitutes that smallest unit of matter is still up for debate, so let’s not worry too much about that and instead just refer to them as “bits”. That is, I’ll define a bit of matter as the smallest unit of matter that cannot be further broken up into smaller pieces. When many bits are put together, they create everything else: from atoms, to oceans, to all living things. From this perspective, to run a faithful simulation of the world would require simulating how every bit of matter in the universe moves around and interacts with other bits across space and time. If we could do this, we would certainly be simulating the universe.
I contend, however, that this sort of “bit-wise” simulation is impossible. Here’s why: To faithfully simulate the activity of every bit of matter in the universe would require a computer that could store at least the same amount of information as there are bits of matter in the universe (for the information theorists, the nomenclature is intentional, but perhaps not precise). But computers are themselves physical things made of matter, and, further, the way that they store information is by — broadly speaking — moving around bits of matter inside of them. This means that to simulate a bit of matter that is somewhere out there in the universe requires at least one additional bit of matter to exist in the computer itself. Thus, a computer can never simulate every bit of matter in the universe because to do so would require that the computer itself be made up of at least as much matter! That is, the computer would have to be at least as massive as the universe in which it exists, which is surely impossible.
A small-scale example may help to spell out this reasoning more clearly. Imagine a pretend universe that consists of only 1,000 bits of matter. The goal is to build a computer inside this universe that can simulate what happens to each of those 1,000 bits over time. But the catch is that the only physical resources available to you inside the universe are those same 1,000 bits of matter. Thus, you are faced with the impossible task of building a computer out of the very same material that you want to simulate. A brilliant engineer might find a way to construct a computer that can represent the 1,000 bits of matter, but simulating how those bits interact over time and space requires additional computing power, and, therefore, additional matter that is not available. Thus, it is impossible to faithfully simulate a 1,000-bit universe from within the universe itself because there is simply not enough matter available to get the job done.
In the case of reality, the problem is magnified. We humans only have access to a minuscule fraction of all the matter that exists in the real universe. Even if we were to somehow build amazingly sophisticated computers, we would never be even remotely close to being able to faithfully simulate every bit of matter in our world.
What about compression?
Clearly, the simulation theorists must concede that the sorts of simulations that they are talking about cannot be so sophisticated that they perfectly replicate the world in which the simulation is being run. Still, it is indisputably possible to simulate simplified versions of any universe. For example, if bits of matter were to always show up in groups of 10, you might be able to get away with simulating how these groups behave instead of how each of the constituent bits does. In this case, you’d only need about 10% of the matter in the universe to build a computer that could represent all the groups of bits in the universe. Thus, the simulation theorist might argue, simplified simulations may be a real possibility.
It is important to recognize, however, that simplified simulations necessitate that a simulated universe is always less complex (i.e. contains less information) than the universe in which that simulation is being conducted. While a simulation that looks at groups of 10 bits of matter at a time may closely mirror reality, it is necessarily a simplified version of the real world and always leaves something out. Moreover, practical constraints on the size of a computer that we could possibly build in even our most fantastical dreams stipulate that any simulation that we could ever run would necessarily involve only an overwhelmingly simplified version of the real universe.
It is in the face of this reasoning that the simulation hypothesis breaks down completely. Recall that the original argument for the theory suggests that there are countless simulated universes, each nested one inside the next with only one base reality at the top. The reason for assuming that this many simulations might exist relies on an inferred similarity between the people that live in each of the simulated worlds. That is, the simulation hypothesis critically depends on the assumption that people in each of the simulated worlds are at least psychologically similar to humans in base reality in that they are all themselves motivated enough to want to run simulations. But, as we’ve just seen, we have reason to assume that any simulated universe is in fact quite different from the universe in which that simulation is run because there must be drastic simplification between the two. Thus, there is no clear reason to assume that the people that live in any simulated world would ever share the same idiosyncratic psychological motivations that inspire those in base reality to run simulations.
But that’s not all. There’s even some evidence to think that simplified versions of human beings might not ever share the desire to run simulations. Chimpanzees are psychologically similar to human beings. These animals, however, do not express a clear desire to run complex simulations of the world. Perhaps with more evolutionary time chimpanzees could develop into a species capable of running such simulations, but there is still no reason to assume that they will develop the highly peculiar psychological trait that makes doing so interesting or worthwhile. If those species most similar to humans in this world don't share a desire to run simulations, why should we assume that people living in a nested chain of increasingly simplified universes ever would?
He’s Wrong
At its crux, Musk’s version of the simulation hypothesis fails because it critically relies on an inferred similarity between base reality and simulated universes; and, as we’ve seen, there is no good reason to suppose that this similarity exists. Unfortunately, this doesn’t leave us much of anywhere: the universe is still an immensely confusing and mysterious place and what we know about how it works or what we’re doing here remains embarrassingly sparse. But, at least the next time your Tesla-Loving, Stonk-Tracking, Billionaire-Worshipping friend tries to tell you how Elon the Profit has proclaimed that this is all just one big video game, you’ll have something to say.
