 |
| Are we reality or a computation? The simulation hypothesis at the edge of modern physics. |
There is a question you can reach from two completely different directions. You can start in philosophy — in Descartes's demon, in Plato's cave, in Zhuangzi's butterfly dream — and find yourself asking whether the world you perceive is real. Or you can start in physics — in the discrete structure of quantum mechanics, in the mathematical precision of the laws of nature, in the computational architecture of spacetime itself — and find yourself asking the same question in an entirely different register. The simulation hypothesis is the place where these two paths converge. It is often dismissed as science fiction dressed in academic clothing. It is also, depending on which version you take seriously, either a logical trap that forces an uncomfortable conclusion or a genuinely testable claim about the physical structure of the universe. This final episode of the series is not going to tell you whether we live in a simulation. That would be easy to write and dishonest to read. What it is going to do is show you why the question is harder than you think, why the smartest people who have examined it land in very different places, and why the answer — whatever it is — may say something profound about the relationship between mathematics, physics, and the nature of existence itself.
Limits of the Universe · Episode 05 of 05 · Series Finale
Are We Simulated?
If the universe is mathematics, and mathematics exists independently of matter, then reality may be a computation. The simulation hypothesis is no longer just philosophy — it has become a question that physics must answer.
Bostrom · Tegmark · Hossenfelder · Aaronson · Gates · Where the argument stands in 2025
Series Complete — All Five Episodes
EP 05
Are We Simulated? — The Simulation Hypothesis
YOU ARE HERE
From Plato to Descartes: The Ancient Version of the Question
Around 380 BCE, Plato described prisoners chained in a cave, facing a wall. Behind them, puppets cast shadows on that wall, and the prisoners, having never seen anything else, mistake the shadows for reality. It is one of philosophy's most enduring thought experiments — not because it suggests that the world is literally a projection of something behind it, but because it raises a question that has no clean answer: if your entire perceptual apparatus is pointed at shadows, what would it even mean to look behind you?
In China, around the same era, the Daoist philosopher Zhuangzi wrote of dreaming he was a butterfly and waking as a man — and genuinely not knowing, upon waking, whether he was a man who had dreamed of being a butterfly or a butterfly now dreaming of being a man. René Descartes, in 1641, pressed the same anxiety into formal philosophical structure: his Meditations on First Philosophy asked whether an evil demon of supreme power might be manipulating his senses so completely that nothing he perceived corresponded to reality.
These are the ancestors of the simulation hypothesis. But they differ from its modern form in a crucial way. Plato, Zhuangzi, and Descartes were raising epistemological questions — questions about what we can know. The modern simulation hypothesis, at least in its more serious formulations, makes a different kind of claim: not merely that we might be deceived about reality, but that the structure of physics itself provides evidence for or against the proposition. It has moved from philosophy into the territory where physics has opinions.
Bostrom's Trilemma: A Logical Trap With No Good Exits
In 2003, philosopher Nick Bostrom of Oxford published a paper titled "Are You Living in a Computer Simulation?" in the Philosophical Quarterly. The paper was not, despite its title, an argument that we are in a simulation. It was a trilemma — a demonstration that one of three propositions must almost certainly be true, with no comfortable way to avoid all three.
The three propositions are these. First: the fraction of civilisations like ours that survive long enough to reach what Bostrom calls the "posthuman" stage — possessing the computational resources to simulate entire ancestral civilisations — is very close to zero. Something, a filter of some kind, prevents most civilisations from ever reaching that technological threshold. Second: posthuman civilisations overwhelmingly choose not to run ancestor simulations, perhaps for ethical reasons, perhaps out of indifference, perhaps because of constraints we cannot currently anticipate. Third: we are almost certainly living in a simulation.
The logic binding these three is probability. If proposition one and proposition two are both false — if civilisations routinely survive to the posthuman stage and routinely run ancestor simulations — then the number of simulated minds would enormously exceed the number of minds in base reality. There would be billions of simulated civilisations for every real one. In that case, any randomly selected mind is overwhelmingly likely to be a simulated mind. You, reading this, would almost certainly be simulated.
This is not a claim that simulation is true. It is a claim that you must accept one of three deeply uncomfortable things: universal civilisational extinction before the posthuman stage; universal civilisational abstinence from ancestor simulation once that stage is reached; or we are in a simulation. Bostrom himself assigns roughly 20% probability to the simulation hypothesis — less than even odds across the three propositions — but he does not claim to know which is true. The trilemma forces the choice. It does not make the choice.
Bostrom's Trilemma — simplified: Either (1) almost all civilisations go extinct before they can simulate, or (2) almost all posthuman civilisations choose never to simulate, or (3) we are almost certainly living in a simulation. At least one of these is almost certainly true. None of them is comfortable.
The Numbers: Why the Trilemma Forces a Disturbing Conclusion
To understand why the probabilistic conclusion is so strong, consider what a posthuman civilisation would be capable of. A civilisation that has been technologically advanced for even a few thousand years — a trivially short time on astronomical scales — might have access to computing resources that dwarf anything we can currently build. Bostrom references estimates from theoretical physics suggesting that the computing resources of a single planetary mass converted to computing substrate would be sufficient to run simulations of the entire history of Earth's civilisation many billions of times over.
If even a small fraction of posthuman civilisations run such simulations — even one in a million — and each of those simulations is run even once, the number of simulated minds in the universe exceeds the number of "real" minds by many orders of magnitude. The simulated minds, from the inside, would be indistinguishable from real minds. They would have memories, experiences, perceptions, and thoughts. There would be billions of simulated versions of you for every real version. Under any standard application of probabilistic reasoning about your own location in the space of all minds, you should expect to be simulated.
The only escape from this conclusion is to accept that proposition one or proposition two is nearly universally true. Proposition one is the more sobering option: a Great Filter ahead of us, some catastrophic barrier — artificial superintelligence, engineered pandemic, nuclear war, climate collapse — that kills essentially all civilisations before they reach the posthuman stage. The Fermi paradox — the puzzling silence of the universe, which should by now contain abundant evidence of technologically advanced civilisations — is sometimes cited as indirect evidence that such a filter exists. If it does, the absence of simulations is explained. But so is our likely extinction.
From Philosophy to Physics: The Mathematical Universe
Bostrom's argument is statistical. It does not engage with the physics of what a simulation would require or what evidence might distinguish a simulated universe from a non-simulated one. A separate line of thinking, rooted in physics and the philosophy of mathematics, approaches the simulation question from a different direction entirely.
The observation is this: the universe, at every scale at which we have been able to examine it, appears to be mathematical. Not merely described by mathematics — actually is mathematics. The laws of physics are equations. The elementary particles are characterised entirely by numbers — mass, charge, spin. The structure of spacetime is a mathematical manifold. When you strip away everything that is not essential to describing an electron's behaviour, what remains is a set of quantum numbers and transformation rules under symmetry operations. There is no residue, no leftover "stuff" that is not mathematical. The electron, in the fullest physical description available to us, is the mathematics.
This observation was noted by the physicist Eugene Wigner in 1960 in a famous essay titled "The Unreasonable Effectiveness of Mathematics in the Natural Sciences." Wigner was struck by the fact that mathematical structures developed by pure mathematicians for entirely abstract reasons — with no application in mind, no physical motivation — repeatedly turn out to describe physical reality with extraordinary accuracy. Complex numbers were invented to solve polynomial equations; they became essential for quantum mechanics. Non-Euclidean geometry was a mathematical curiosity; Einstein found it was the geometry of spacetime. Group theory was pure algebra; it turned out to organise all known fundamental particles. The correspondence between abstract mathematics and physical reality is too strong, too consistent, and too deep to be a coincidence — though what exactly it signifies remains philosophically contested.
Max Tegmark and the Mathematical Universe Hypothesis
MIT cosmologist Max Tegmark takes Wigner's observation to its logical extreme. In his Mathematical Universe Hypothesis (MUH), Tegmark proposes that the universe does not merely follow mathematical laws — it literally is a mathematical structure. Physical existence and mathematical existence are the same thing. There is no "physical stuff" separate from the mathematics; the mathematics is all there is. Every consistent mathematical structure that exists abstractly is, in Tegmark's framework, also a physical universe.
This has a direct bearing on the simulation question. If the universe is a mathematical structure, and mathematical structures can be computed, then a sufficiently powerful computer running the relevant mathematics would not be simulating the universe — it would be instantiating it. There would be no meaningful distinction between the mathematics running on a computer and the "real" universe. As Tegmark noted: if he were a character in a video game and began to notice that the rules of his world were entirely mathematical, that is precisely what he would expect to find. The video game character and the physicist in a physical universe would be epistemically indistinguishable.
The Mathematical Universe Hypothesis has been criticised on multiple grounds. The most fundamental objection is that it is unfalsifiable — it makes no prediction that differs from predictions made by more conventional physical theories. If every consistent mathematical structure is a universe, the MUH predicts everything and therefore constrains nothing. A theory that is consistent with all possible observations explains none of them. Tegmark's defenders argue that the MUH is not a physical theory in the conventional sense but a metaphysical framework — a way of answering the question "why is there something rather than nothing?" not by invoking a creator but by arguing that mathematical structures exist necessarily, and therefore physical reality (being a mathematical structure) exists necessarily too.
Error-Correcting Codes in the Equations of Physics
In 2010, theoretical physicist S. James Gates Jr. at the University of Maryland reported something that he described as personally unsettling. Working on the mathematics of supersymmetry — the theory proposing that every known particle has a supersymmetric partner — he found that the equations describing the behaviour of quarks, leptons, and the forces between them contained, embedded within their structure, something that looked unmistakably familiar: error-correcting codes.
Error-correcting codes are the mathematical structures that make digital communication reliable. When you load a webpage, the data transmitted over the network includes redundant information — additional bits that allow the receiving system to detect and correct errors introduced by noise. The specific type of codes Gates found in his supersymmetry equations — doubly-even binary self-dual linear block codes — are exactly the codes used in systems like the one that makes your browser work. They are structures deliberately invented by engineers in the twentieth century to solve a practical engineering problem. Gates found them embedded in equations describing the fundamental physics of particles and forces, equations whose forms are constrained entirely by the symmetries of nature.
Gates's response to this discovery, in a widely circulated account, was: "That's what brought me to this very stark realisation that I could no longer say that people like Max Tegmark are crazy." He was careful to note that finding such codes in the equations of physics does not prove simulation — these are equations describing an abstract symmetry structure, and finding elegant mathematical patterns in them is not necessarily surprising. But the specificity of the match — that it is precisely the structure of error-correcting codes, not merely some vague analogy to information processing — gave him pause. It gave a number of physicists pause.
The counter-argument is that error-correcting codes are, at a deep level, just particular kinds of mathematical structure — and given the richness of mathematics, finding specific mathematical structures appearing in physical equations is not as surprising as it sounds. The universe is not the only place error-correcting codes appear. They appear throughout pure mathematics as well. Correlation is not causation; appearing in physics equations does not mean physics is computation.
Quantum Mechanics: Discreteness, Superposition, and the Limits of Reality
Several features of quantum mechanics have been cited as suggestive of computational or simulation-like structure, though each requires careful interpretation. The most commonly cited is discreteness. Physical quantities in quantum mechanics — energy levels, angular momentum, electric charge — come in discrete units. You cannot have half a unit of electric charge or 1.7 units of angular momentum. The universe quantises these quantities. Computers, by their nature, process discrete information. The argument runs: if reality is continuous, simulation would be harder; if reality is discrete, it looks more like something that could be computed.
The second feature is the observer-dependence of quantum states. Before measurement, a quantum particle exists in a superposition of multiple possible states. Measurement collapses this superposition to a single definite value. Some have suggested this resembles the "rendering" of a video game — where distant, unobserved regions of the map are not rendered in full resolution until a player approaches and observation becomes necessary. The universe, on this reading, only "calculates" what is being observed, conserving computational resources. This is a suggestive metaphor. It is not a scientific argument. Quantum mechanics does not say unobserved regions are not "real" — the wavefunction evolves according to the Schrödinger equation whether observed or not.
The Planck length — approximately 1.616 × 10−35 metres, the smallest length at which current physics makes sense — has been cited as a potential "pixel size" of the universe. Below the Planck scale, the concepts of space and time as continuous entities break down, and quantum gravitational effects dominate. This discreteness at the smallest scales is consistent with, though far from proof of, a computational substrate. Many quantum gravity theories — loop quantum gravity, spin foam models — propose that spacetime is fundamentally discrete at the Planck scale for reasons having nothing to do with simulation.
Physical Features Cited in Simulation Arguments
| Feature |
Simulation-suggestive reading |
Alternative explanation |
| Discrete quantum states | Resembles digital computation | Intrinsic to quantum mechanics |
| Planck-scale discreteness | Possible "pixel size" of reality | Predicted by quantum gravity theories |
| Error-correcting codes in supersymmetry equations | Structure of computation embedded in physics | Rich mathematics produces many structures |
| Observer-dependent wavefunction collapse | Universe "renders" only what is observed | Wavefunction evolves continuously without observation |
| Fine-tuned physical constants | Parameters set by a programmer | Anthropic selection; multiverse theories |
The Case Against: Hossenfelder, Lorentz Invariance, and Computational Limits
Theoretical physicist Sabine Hossenfelder has made the most direct and technically grounded case against the simulation hypothesis from within physics. Her argument is not philosophical scepticism; it is a specific physics objection. The universe, as far as all current measurements can determine, is Lorentz-invariant — the laws of physics look the same in all inertial reference frames, regardless of speed or direction of motion. This is one of the most precisely tested symmetries in physics.
Any simulation running on a discrete computational lattice — the obvious way to implement a finite computation over a finite space — would break Lorentz invariance. A lattice has preferred directions: the directions along its axes. A particle propagating at an angle to the lattice axes would experience slightly different physics than one moving parallel to them. This anisotropy would show up in high-energy cosmic ray physics. Experiments searching for Lorentz invariance violation in cosmic rays — at energies many orders of magnitude above what any accelerator can produce — have found none. The universe behaves as if it is perfectly continuous and Lorentz-invariant at all measurable scales.
Hossenfelder's argument is not that a simulation is impossible in principle — a sufficiently clever simulation might use different computational architecture, or might use continuous rather than discrete mathematics. Her argument is that the naive, most natural version of a computational simulation — a discrete lattice — is essentially ruled out by current physics. More exotic simulations remain possible but require additional assumptions that have no independent support.
A separate objection concerns computational resources. Seth Lloyd of MIT has estimated the maximum computational capacity of the observable universe — treating every particle in the universe as a computing element operating at the maximum rate permitted by the uncertainty principle, for the entire age of the universe. The result is approximately 10120 operations. This is a very large number. It is not, however, infinite. And the information content of a simulation of the observable universe, at quantum resolution, is also approximately 10120 bits. A simulation of our universe, run at the same resolution as our universe, would require a computer at least as large and as complex as our universe. The simulation does not explain the universe; it merely posits another universe to house the computer.
Scott Aaronson's Position: The Branch Where the Argument Saws Itself Off
Computer scientist Scott Aaronson of the University of Texas, one of the world's leading experts on computational complexity and quantum computing, has engaged with the simulation hypothesis at length and arrived at a nuanced position that neither fully endorses nor dismisses it. His most significant objection targets the probabilistic foundation of Bostrom's argument.
Aaronson's concern is what he describes as the argument sawing off the branch it is sitting on. The trilemma rests on reasoning about the future trajectory of our civilisation — about the technological development of posthuman descendants and their computational capabilities. But that reasoning depends on the accumulated knowledge of science and history. If we are already in a simulation, Aaronson points out, there is no guarantee that our knowledge of science and history — the very foundation on which the trilemma's reasoning rests — corresponds to anything in the base reality running the simulation. The simulation's programmers could have seeded us with false memories, false historical records, false scientific observations. The argument uses simulated knowledge to conclude that we might be simulated. This is not a logical refutation, but it is a serious weakening of the argument's epistemic foundation.
Aaronson also notes a structural problem: the simulations our posthuman descendants would run would themselves have to run on computers in our universe. Those computers fit inside our universe. The simulated universes they model would therefore, in some meaningful sense, be smaller than our universe in terms of the number of bits required to describe them. The simulation is not explaining our universe; it is describing a different, smaller computation nested inside our universe. Each level of simulation runs on hardware from the level above, meaning base reality — whatever it is — must be at least as computationally rich as the most complex simulation it spawns. The regress does not dissolve the problem of what base reality is.
The Deepest Problem: Can the Question Even Be Answered?
The most fundamental difficulty with the simulation hypothesis is not that the evidence is weak — it is that any evidence that could support it could also be explained without it. A discrete Planck-scale structure to spacetime is predicted by quantum gravity theories. Error-correcting codes in supersymmetry equations could be the result of the mathematical richness of symmetry structures. The fine-tuning of physical constants could reflect anthropic selection in a multiverse. Every observation cited as suggestive of simulation has an alternative explanation within conventional physics.
This is not merely a practical problem of insufficient evidence. It may be a structural impossibility. Any universe — simulated or not — that contains beings capable of doing physics would appear, from the inside, to have consistent mathematical laws. The laws would look as though they were designed, because the beings doing the looking are themselves the product of those laws. There may be no observation that a being inside a simulation could make that would distinguish their situation from being in a base reality. If that is true — and it is far from certain, but it is a serious possibility — then the question "are we simulated?" is not merely unanswered but unanswerable in principle. It would join a small class of questions that are logically coherent but lie permanently outside the reach of empirical science.
Some physicists, including Hossenfelder, go further and classify the simulation hypothesis as pseudoscience on these grounds: a hypothesis that cannot, even in principle, make a prediction that differs from competing hypotheses is scientifically sterile. It cannot be tested, it cannot be confirmed, and it cannot be falsified. By Karl Popper's criterion of falsifiability, it does not qualify as a scientific hypothesis at all. It belongs, in this view, with the Cartesian demon and Zhuangzi's butterfly — as a thought experiment of genuine philosophical interest but no empirical purchase.
Others disagree. There are proposals — not yet experimentally realised, but logically coherent — for tests that would look for lattice-like artefacts in the angular distribution of ultra-high-energy cosmic rays, which a discrete computational substrate might produce. The Lorentz invariance constraint Hossenfelder cites is already one such test, and the universe has passed it at the scales we can probe. Whether it would pass it at Planck-scale resolution is unknown — and unknowable with current technology. Future experiments may or may not close that gap.
What the Series Has Shown: The Shape of What We Don't Know
This series has moved through five problems at the outer limits of what physics can currently answer. Each episode arrived at the same basic conclusion by a different route.
At absolute zero, atoms lose their individuality and become a single quantum entity — which is not merely strange but deeply incompatible with our ordinary intuitions about what a "thing" is. In the invisible universe, 95% of the cosmos is made of something we cannot identify after three decades of searching with the most sensitive instruments ever built. At the event horizon, two perfectly verified physical theories make contradictory predictions about what happens to information, and the mathematics that resolves the contradiction still does not explain the physics. In the cosmic microwave background, the oldest light in the universe carries anomalies at the largest scales that the standard model cannot account for, and two independent measurements of the expansion rate of the universe disagree at 5 sigma. And in the simulation question, the deepest framework we can construct for asking what reality is may be unanswerable in principle.
The pattern is not one of failure. It is the pattern of a field working at the edge of what it currently understands. Each of these problems exists because the instruments and theories powerful enough to expose the question were not available before. Absolute zero was approachable only with laser cooling and magnetic traps. The dark universe was revealed only by the precision of modern spectroscopy and gravitational lensing. The information paradox sharpened into crisis only when string theory and quantum information theory became precise enough to reveal the internal contradiction. The Hubble tension emerged only when multiple measurement chains became accurate enough that their disagreement could not be attributed to error. The simulation hypothesis became a serious question only when computing power and the mathematics of information theory provided a concrete enough framework to ask it rigorously.
The limits of the universe, as they stand in 2025, are not the ends of understanding. They are its growing edge. Each of these five problems has been worked on for decades by some of the most capable scientific minds in history, and each remains genuinely, substantially open. Not because the questions are too hard to answer, but because answering them correctly will require ideas that do not yet exist.
Whether we are simulated or not, we are, at minimum, the part of the universe that is capable of asking the question. That is a strange and remarkable thing to be — regardless of what substrate the asking is running on.
Limits of the Universe — Series Complete
Five Episodes. Five Open Questions. Zero Easy Answers.
Absolute Zero · Dark Universe · Event Horizon · CMB · Simulation
Disclaimer: This article is written for general educational purposes. The simulation hypothesis is not an established scientific theory; it is an active area of philosophical and theoretical inquiry. Nick Bostrom's 2003 trilemma is accurately represented as a logical argument, not as a scientific claim. Max Tegmark's Mathematical Universe Hypothesis is a philosophical position held by a minority of physicists; it is not scientific consensus. S. James Gates's observation of error-correcting codes in supersymmetry equations is a real mathematical result, accurately described here. Sabine Hossenfelder's Lorentz invariance argument is a genuine physics objection, accurately represented. Scott Aaronson's position is drawn from his publicly available 2024 blog discussion. No empirical evidence currently confirms or definitively refutes the simulation hypothesis. All sources are publicly available and legally accessible.
References
- Bostrom, N. (2003). Are you living in a computer simulation? Philosophical Quarterly, 53(211), 243–255. → simulation-argument.com/simulation.pdf
- Tegmark, M. (2014). Our Mathematical Universe: My Quest for the Ultimate Nature of Reality. Deckle Edge. [Book — foundational text for the Mathematical Universe Hypothesis]
- Wigner, E. (1960). The unreasonable effectiveness of mathematics in the natural sciences. Communications on Pure and Applied Mathematics, 13, 1–14. → doi.org/10.1002/cpa.3160130102
- Gates, S.J. (2010). Symbols of power. Physics World, 23(6), 34–39. [On error-correcting codes in supersymmetry equations]
- Hossenfelder, S. (2021). The simulation hypothesis is pseudoscience. Backreaction blog. → backreaction.blogspot.com
- Hossenfelder, S. (2017). No, we probably don't live in a computer simulation. Backreaction blog. → backreaction.blogspot.com
- Aaronson, S. (2024). On whether we're living in a simulation. Shtetl-Optimized blog. → scottaaronson.blog/?p=7774
- Lloyd, S. (2002). Computational capacity of the universe. Physical Review Letters, 88, 237901. → doi.org/10.1103/PhysRevLett.88.237901
- Wikipedia. Simulation hypothesis. (Regularly updated.) → en.wikipedia.org/wiki/Simulation_hypothesis
The cosmos is not merely a vast, empty void; it is a laboratory where the fundamental laws of reality are pushed to their breaking point. When we look at the night sky, we aren't just seeing stars—we are looking at history. We are seeing light that has travelled for billions of years, witnessing the expansion of space-time, and grappling with the fact that 95% of everything out there is completely invisible to us.
In this section of Decoding Curiosity, we explore the outer reaches of understanding—from the faint echoes of the Big Bang to the crushing silence of black holes. This isn't just a collection of facts about planets and galaxies; it is a journey into the mechanics of existence, where mathematics and curiosity meet the infinite.
Explore all our articles on the universe here: Explore the Space Collection
Comments
Post a Comment