File AN-R08 · Anomalies · Fact recordFiled 2026-09-17

Are we living in a simulation? The argument and its critics

Nobody knows, and no observation so far supports the idea. The simulation hypothesis in its modern form comes from Nick Bostrom's 2003 argument, which does not say we are simulated: it says that almost no civilization like ours reaches the needed technology, or almost none uses it to simulate its ancestors, or we are almost certainly simulated. Physicists and philosophers still dispute its premises, its probabilities and whether it can be tested at all.

What the simulation hypothesis is

The simulation hypothesis is the idea that the world we experience, our minds included, runs inside a computer built by a more advanced civilization, as in the Matrix films. Are we living in a simulation? Nobody knows. No observation has produced evidence for it, and the arguments on both sides are philosophical or mathematical, not measurements.

The modern version comes from the philosopher Nick Bostrom, whose paper "Are You Living in a Computer Simulation?" was published in Philosophical Quarterly in 2003. It does not claim we are simulated. It argues that at least one of three things is true: almost no civilization like ours reaches a technologically mature, posthuman stage; almost no posthuman civilization runs many simulations of its ancestors; or almost everyone with experiences like ours is simulated.

How the argument works

Explained step by step, it rests on two premises. One, which Bostrom concedes is contested, is that a computer copying a brain's processes down to individual synapses would be conscious. The other is cost. He puts a simulation of humankind's entire mental history at about 10³³ to 10³⁶ operations, which a planet-mass computer doing 10⁴² operations per second could run with less than a millionth of its capacity for one second.

If mature civilizations ran many such simulations, simulated minds would vastly outnumber biological ones, and unless you have reason to think your experience is unusual, your confidence that you are simulated should match that fraction. Bostrom's own verdict was restrained: given our ignorance, belief should be spread roughly evenly across the three options. Unless we are simulated now, he concluded, our descendants will almost certainly never run such simulations.

The popular claim

In public retellings, the odds tend to grow. At the American Museum of Natural History's 2016 Isaac Asimov Memorial Debate, the astrophysicist Neil deGrasse Tyson put the chance that our existence is someone else's computer program at 50-50. Elon Musk told a 2016 conference that the odds we live in base reality are one in billions.

The examples offered as evidence for the simulation hypothesis are features of physics that resemble software. At that debate, the MIT cosmologist Max Tegmark said a computer-game character would eventually find its rules rigid and mathematical, as physicists find ours. James Gates of the University of Maryland described finding error-correcting codes, the kind browsers rely on, in equations about quarks, electrons and supersymmetry. Tyson has relished the thought that the speed of light is a limit a programmer set.

The scientific explanation

At the same debate, the Harvard physicist Lisa Randall put the chance at effectively zero, saying the argument is not based on well-defined probabilities. Nor do the cited features tell the cases apart: an unsimulated universe could have mathematical laws and a speed limit too. In 2021 the physicist Sabine Hossenfelder called the hypothesis pseudoscience, since nobody currently knows how to reproduce general relativity and the Standard Model of particle physics with a computer algorithm.

Algorithmic versions of natural laws, she wrote, are usually incompatible with the symmetries of Einstein's relativity, and proponents assume far more than they explain. Her charge is not that the idea is wrong but that believing it takes faith, not logic. Bostrom's paper also leaves room to explain away any anomaly: a simulator could fill in microscopic detail only when someone looks, edit the memories of anyone who noticed a problem, or rerun the last few seconds.

Can it be tested?

Only narrow versions can. In 2012 the physicists Silas Beane, Zohreh Davoudi and Martin Savage imagined a simulation run on a cubic grid of space-time points, like today's lattice calculations of the strong nuclear force. Such a grid could make the highest-energy cosmic rays break rotational symmetry in a pattern reflecting its structure. From the cosmic-ray energy cutoff they derived a bound on how fine the grid must be: a limit, not a detection.

A different design would leave different traces, or none. The New York University philosopher David Chalmers said at the 2016 debate that we will never get proof we are not simulated, since any evidence could itself be simulated. David Kipping, a Columbia University astronomer, has called the question arguably not testable, and asked how an unfalsifiable idea can count as science.

What is still debated

The probabilities are contested too. In a 2020 paper in the journal Universe, Kipping merged Bostrom's first two options, since both mean no simulations, and gave that and the simulation scenario equal starting odds. His Bayesian analysis put the chance that we are simulated just under 50 percent. If humans ever build simulations containing conscious beings, he found, the odds would swing sharply toward our being simulated.

Bostrom called the equal starting odds shaky: applied to his three options, the same move gives each a third, and dividing the possibilities differently gives any answer one likes. The philosopher Brian Weatherson, then at Brown University, attacked the step from fractions to personal belief, arguing that each of four readings of Bostrom's indifference principle either fails or needs an assumption we have no reason to accept: that none of our evidence bears on whether we are simulated.

Claims that it has been disproved

Claimed disproofs have outrun the evidence too. A 2017 University of Oxford release on a Science Advances paper by Zohar Ringel and Dmitry Kovrizhin opened with the simulation question. The paper argued that quantum Monte Carlo, a broad class of classical algorithms, fails for certain quantum systems because of the sign problem. That limits one family of algorithms, not every possible computer, and the release itself said only that the work may reassure some people.

In October 2025, UBC Okanagan announced that Mir Faizal and colleagues, including Lawrence Krauss, had mathematically proven the universe cannot be a simulation. Their paper in the Journal of Holography Applications in Physics uses Gödel's incompleteness theorem and related results to argue that a complete description of reality needs non-algorithmic understanding, which no simulation could have. It is a mathematical argument, not an observation, and only as strong as that premise.

Verdict

Is the simulation hypothesis real? It is a serious, carefully argued hypothesis with no evidence that it is true, and Real or Legend has filed it as disputed. Bostrom's argument is a three-way choice, not a discovery. The popular examples fit an unsimulated universe equally well, the one concrete test covers a single design, and claimed disproofs rest on premises of their own. Views still run from one in billions for base reality to effectively zero for a simulation.