The Zero-Player Game

In 1970 the British mathematician John Conway invented something he called the Game of Life, and then spent years mildly annoyed that it became the most famous thing he ever made. It is barely a game. You do not play it. You set up a grid of cells, switch a few of them on, and walk away. From then on it runs itself, with no players and almost no rules: a living cell with two or three living neighbours survives, a dead cell with exactly three neighbours comes alive, and everything else dies. That is the entire universe of the Game of Life.

What happens next is the part that has kept people up at night for fifty years. Out of those four lines of rules, structures appear. Little shapes called gliders crawl across the grid forever. Patterns called guns fire off an endless stream of them. People have built clocks inside it, then logic gates, then, eventually, a working computer that can run the Game of Life inside the Game of Life. A few switched-on cells and some arithmetic, and you get open-ended complexity that nobody designed and nobody can fully predict.

Sit with that long enough and a vertigo sets in. If a simulation this trivial, with this little input, can grow something that looks alive, what happens when you scale the idea up? Not dots on a grid, but galaxies, planets, weather, minds. And if a rich enough universe can in principle be computed, then the unsettling question is not whether someone could build one. It is how we could ever know we are not already inside one.

Do not take my word for it. Press play, draw some cells, and watch complexity boot up from almost nothing. This is the whole universe of the Game of Life, running right here in your browser.

Play Step Random Glider gun Clear gen 0

The argument is older and sharper than it sounds

The intuition is fun, but on its own it is just a vibe. The reason the simulation hypothesis is taken seriously in philosophy departments and not just on late-night forums is a single tight argument made in 2003 by the Oxford philosopher Nick Bostrom. It is worth getting right, because almost everyone repeats a sloppy version of it.

Bostrom does not claim we are in a simulation. He claims that at least one of three statements must be true, and that is a much harder thing to escape. Either, first, civilizations almost always wipe themselves out before they get powerful enough to run huge, realistic simulations of conscious beings. Or, second, civilizations that do reach that power almost never bother running such simulations. Or, third, we are almost certainly living in one. The logic is a counting argument. If even a handful of advanced civilizations ever run many high-fidelity ancestor simulations, then simulated minds vastly outnumber original biological ones, and a randomly selected mind, yours, should expect to be one of the many rather than one of the few.

Notice what this does. To feel safe, you have to bet that advanced civilizations either reliably self-destruct or reliably lose all interest in re-running their own history, every single time, across the whole cosmos and all of future time. Those are not crazy bets. But they are bets, and most people who hear the argument realize they cannot comfortably make either one. That is the trap. The trilemma converts a wild idea into an uncomfortable process of elimination.

Why a fake universe would be cheaper than you think

The usual objection is scale. Simulating a universe with hundreds of billions of galaxies, atom by atom, sounds absurdly expensive. But that objection quietly assumes the simulators would be as naive as a first-year programmer, rendering everything all the time. No working simulation does that, and ours, if it is one, clearly would not either.

Start with a fact that should bother you more than it does. You have never actually seen an atom. Nobody has, not directly. The chair you are sitting on, the screen you are reading, your own hand, all of it is, as far as you can ever verify, a model assembled in your head from a thin trickle of sensory data. Everything you call the world is your subjective reality, reconstructed behind your eyes. There is no way to step outside your own mind and confirm an objective world that exists in full detail whether or not anyone observes it. To prove something exists unobserved, you would have to observe it unobserved, which is a contradiction.

A clever simulator would exploit exactly this. It would not need to compute a universe. It would need to compute the experience of one, and only the parts under observation. This is precisely how video games already work. They do not render the room behind you, or the inside of a wall, or a distant city, until your viewpoint demands it. They use tricks called culling and level of detail to keep the machine cheap, drawing the world into existence just ahead of where you look and letting it dissolve behind you. The famous riddle about a tree falling in an empty forest stops being a riddle and becomes an optimization: if no observer is near, there is no reason to compute the fall at all. You arrive later and find a fallen tree, and the simulation has spent nothing on the event itself.

Push this further and the cost collapses again. If reality is reconstructed inside minds, the simulators might not need to render a shared world at all. They might only need to simulate the minds, and feed each one a consistent stream. Every other person you have ever met could, in principle, be part of the same process, and you would have no test to tell the difference. That is a deeply arrogant and lonely thought, and I do not believe it. But the point of the argument is not that it is true. The point is that you cannot prove it false, which is exactly the property a good simulation would have.

The “evidence,” and why I would not bet the house on it

Once you are primed to see a simulation, the universe starts to look suspiciously like a computer struggling with its own settings. This is the fun part, and also the part where you should keep one hand on your wallet.

There is the speed of light, a hard cosmic limit on how fast anything can happen, which looks a lot like the maximum clock speed of whatever is running the show. There is the Heisenberg uncertainty principle, the rule that you cannot pin down both a particle’s position and its momentum at once, which looks like a resolution limit, a refusal to compute detail finer than the simulation bothers to store. There is the fact that space and time appear to come in smallest possible chunks at the Planck scale, which looks uncomfortably like pixels and frames. And there is the strange behavior of matter near black holes, where time itself slows to a crawl, which looks like a machine throttling the most information-dense region in the scene to keep from overheating, the cosmic equivalent of a laptop fan spinning up when you open too many tabs.

The most genuinely interesting modern attempt to make this testable comes from the physicist Melvin Vopson, who treats information as a physical thing on par with mass and energy. His proposed “second law of infodynamics” predicts that the information content of isolated systems tends to fall over time, the mirror image of the usual march toward disorder, and he argues this is the kind of behavior you would expect if reality were running on something like data compression. Whether or not it holds up, it is notable simply for being a claim you could in principle check, which is something the simulation idea has almost always lacked.

To actually build such a thing you would need staggering amounts of energy, which is where ideas like the Matryoshka brain come in: nested shells wrapped around a star, each capturing the waste heat of the one inside, turning a sun into a single colossal computer. A civilization that could build one of those around even a few stars could run an enormous number of universes at once. None of this is buildable today. None of it is physically forbidden either, and “not forbidden, just hard” is the entire premise of the argument.

What it would tidily explain

Part of the theory’s pull is how many loose ends it appears to tie off at once. Why is the sky silent, the Fermi paradox, with no sign of the aliens that should be out there? Maybe because the rest of the cosmos is set dressing, rendered as lights in a telescope but never populated, because the simulation is about us. Why does the universe seem so improbably fine-tuned for the existence of life? Maybe because someone tuned it. Why do so many physical constants sit in the narrow band that allows chemistry and stars? Same answer. The multiverse, too, falls out for free: if you can run one simulation, you can run billions, each with slightly different settings, which is a far cheaper way to get infinite parallel worlds than building infinite actual universes.

I find these satisfying and suspicious in equal measure, and for the same reason. A theory that explains everything with one move, and forbids nothing, has bought its tidiness at the price of saying very little. Which brings us to the objections, and to the perspective the breathless version of this story always leaves out.

The counterarguments worth taking seriously

The first is the one philosophers of science reach for immediately. A claim that cannot, even in principle, be proven false is not really a scientific theory; it is metaphysics wearing a lab coat. If any possible observation, a clean limit or a messy one, a silent sky or a crowded one, can be folded into “that is just how they built it,” then the idea has stopped making predictions and started absorbing them. Vopson’s work is interesting precisely because it tries to break out of this trap, but most versions of the argument live comfortably and uselessly inside it.

The second objection is computational, and it cuts deep. Simulating quantum mechanics faithfully is, as far as we know, exponentially expensive: the resources needed blow up with the size of the system, which is the whole reason we are building quantum computers instead of just simulating quantum physics on ordinary ones. A simulator would face the same wall. So either the universe is not as finely quantum as it looks, or a perfect simulation of it is far harder than the casual argument assumes, even with a star’s worth of power.

The third is the one I find most worth carrying, and it comes from the philosopher David Chalmers, who in his 2022 book Reality+ flips the entire emotional charge of the question. Suppose we are in a simulation. So what? A simulated chair is still a real chair, Chalmers argues. It is made of bits rather than atoms, but you can still sit in it, it still holds your weight inside the world you inhabit, and the bits are as real as anything else you can touch. On this view, “we live in a simulation” is not a debunking. It is a claim about what the universe is made of at the bottom layer, no more threatening than learning it is made of quantum fields rather than tiny billiard balls. Your love, your grief, your morning coffee: all still real, all still yours. The simulation does not make life fake. It just makes the substrate stranger than you assumed.

“

If we are in a simulation, that does not make any of this less real. It only changes what “real” is made of. A world of bits you cannot leave is just a world.

”

So what do you do with this

Here is where I land, after years of finding this question equal parts thrilling and slippery. The honest verdict is that we cannot know, and may never be able to, unless something like Vopson’s testable physics pans out. The probability is not zero and not one; it is a shrug with good footnotes. Anyone who tells you they are certain either way is selling something.

But the value of a big idea is rarely the answer. It is what the idea does to your attention while you hold it. The simulation hypothesis forces you to notice that your entire universe really is a reconstruction inside one fragile skull, that you have never once touched the world directly, and that the line between “real” and “rendered” is far blurrier than daily life pretends. Whether or not a posthuman civilization is running you on a Matryoshka brain, you are unmistakably running a kind of zero-player game right now: the laws of physics churn forward on their own, indifferent and rule-bound, and your only move, like Conway watching his grid, is to observe the pattern you have been dealt and decide what to make of it.

That is the part Conway’s little toy got right and never stops teaching. Vast, surprising, beautiful complexity does not need a vast or complicated cause. A few rules and a place to run are enough. Maybe a posthuman ran them. Maybe nobody did. Either way, the gliders are crawling, the galaxies are turning, and we are awake inside it for a little while, which is the one part of the game that does not feel simulated at all.