Are We Living in a Simulation? From a Philosophical Argument to Digital Physics

An Old Problem in a New Technological Form

The question of whether the world is as it appears is much older than computers. Descartes considered the possibility of radical deception in the seventeenth century, while modern philosophy contains brain-in-a-vat, dream-world and related skeptical scenarios. The simulation hypothesis gives the problem a more technological mechanism: instead of a deceiving demon, a computer system generates the experienced world. [1][14]

The modern argument has an important twist, however. Bostrom does not merely ask whether a simulation can be imagined. He asks what would follow if future civilizations produced very large numbers of detailed simulations of people like us. His proposal is therefore statistical and anthropic, not merely skeptical. [1][2]

What Does Bostrom’s Argument Actually Say?

Bostrom’s formulation says that at least one of three propositions must be true:

1. civilizations like humanity almost always become extinct before reaching a technologically “posthuman” stage;
2. technologically mature civilizations almost never run large numbers of simulations of their evolutionary ancestors;
3. if the first two propositions are false, then beings with experiences like ours are overwhelmingly simulated and we should take seriously the possibility that we are among them. [1]

The argument is powerful only if several assumptions are accepted. Consciousness must be at least in principle substrate-independent: a sufficiently detailed computational realization of a mind must genuinely experience rather than merely imitate human behavior. Posthuman civilizations must possess sufficient computing resources. They must also want to run vast numbers of these simulations, and we must be justified in treating ourselves as typical members of the relevant class of observers. [1]

This is why a clarification later emphasized by Bostrom himself matters: the simulation argument does not say that we should believe we are simulated. In 2009 he explicitly wrote that the argument only establishes that at least one of the three propositions is true; he even stated that he personally thought we were probably not simulated. [2]

Where Does the Popular Claim “We Are Almost Certainly Simulated” Come From?

The most intuitive version runs like this: imagine one “base” reality in which an advanced civilization arises. If it creates millions of simulated worlds containing billions of conscious persons, simulated persons greatly outnumber unsimulated ones. If we do not know which observer we are, it may seem statistically more likely that we belong to the larger group. [1]

But this is not a measurement. We do not know the true probability that consciousness can be computationally realized, future technological limits, the motivations of mature civilizations, or even the correct reference class of observers. These are model assumptions. Change the assumptions and the conclusion changes.

David Kipping therefore approached the problem in 2020 using Bayesian model averaging. Because the technical feasibility of Bostrom-like simulations has not been demonstrated, he incorporated uncertainty about the simulation model itself. Under his chosen priors and treatment, the probability that we are simulated remains below 50 percent, approaching that limit for an extremely large number of simulations. If humanity actually began producing conscious Bostrom-style simulations, the odds would change dramatically. [3]

Kipping’s result is not a measured “probability of reality” either. It demonstrates how strongly the answer depends on the statistical framework and starting assumptions.

Digital Physics Is a Different Claim

Long before Bostrom, some researchers asked whether nature itself might be computational. Konrad Zuse’s 1969 Rechnender Raum explored the possibility that physical processes could be described by a discrete computational system resembling a cellular automaton. Edward Fredkin later proposed “digital mechanics” and hypothesized that a computational structure might exactly model microscopic physics. [5][6]

In a more recent context, Gerard ’t Hooft has investigated whether quantum mechanics could emerge from deeper deterministic dynamics related to cellular automata. Yet he also emphasizes that a convincing complete model of our universe achieving this has not been produced. [7]

A universe that is a cellular automaton is not the same thing as a universe inside a computer simulation. The first claim may mean that computational structure is itself fundamental physics. It requires no external programmer, server or “base world.” The second claim says that our world is being implemented by something outside it.

“It from Bit”: Is Information More Fundamental Than Matter?

John Archibald Wheeler popularized the phrase “it from bit” around 1989–1990. He used it while exploring the possibility that physical quantities are deeply connected with information, measurement and binary answers. His idea became influential in later discussions of quantum information. [8]

Rolf Landauer emphasized a different but complementary direction: information is physical. Information processing does not float free from the physical world; storing, erasing and manipulating information is implemented in physical systems and subject to thermodynamic constraints. [9]

Popular culture often combines these slogans into a leap: “if reality is information, it must be a computer simulation.” The conclusion does not follow. A book can be described informationally without being a computer simulation; genetic systems contain information without making DNA literally identical to software running on a laptop. An informational description is not evidence for an external computer.

Does the Universe “Compute Itself”?

David Deutsch formulated a physical version of the Church–Turing principle in 1985 and examined universal computation in quantum mechanics. Seth Lloyd in 2002 estimated upper limits on the amount of information and number of elementary operations that the universe could physically have processed over its history. His starting point was that physical systems can also be considered in terms of registering and processing information. [10][11]

This supports the meaningful statement that the universe can, in a formal sense, be treated as a physical information processor. But again: “the universe can be described as computation” ≠ “someone is computing the universe.” The distinction is comparable to the difference between saying that a planet obeys a mathematical equation of motion and saying there must be a computer somewhere calculating its orbit in real time.

Does Quantum Physics Show That Reality Is Pixelated?

A popular claim says that the Planck length, energy quantization or discrete quantum outcomes reveal the “pixels of space.” This is too quick. The Planck length is a natural scale constructed from the gravitational constant, Planck’s constant and the speed of light; by itself it does not imply that space is arranged on a cubic grid. Quantization of particular observables likewise does not imply that all space is a classical digital lattice.

Some approaches to quantum gravity do investigate discrete structures, while others use very different mathematical concepts. We currently have no experimental evidence that spacetime is an ordinary computer lattice with a fixed “pixel” size. [4][7]

Similarities between quantum physics and computational language — quantum information, qubits, discrete measurement outcomes — therefore should not be converted into evidence of simulation. Quantum computers are devices that use physics; their existence does not imply that physics itself is software running on another device.

Could We Experimentally Detect a Simulation?

That depends on what kind of simulation is proposed. If a simulator uses a particular numerical scheme and that scheme leaves physical artifacts, those artifacts could in principle be sought. In 2012 Silas Beane, Zohreh Davoudi and Martin Savage analyzed a very specific scenario: the universe as a numerical simulation on a cubic spacetime lattice analogous to lattices used in quantum chromodynamics calculations. [4]

Such a lattice could produce violations of exact rotational symmetry or other signatures at very high energies. The authors derived a bound on the possible lattice spacing from the high-energy cosmic-ray spectrum and suggested that the distribution of the highest-energy cosmic rays might be searched for a specific lattice-induced breaking of symmetry. [4]

A negative result would not refute “simulation” in general. It would rule out or constrain only that family of simulations — for example, a cubic lattice with particular numerical properties. A more sophisticated simulator might not use such a lattice or could push its artifacts below measurable scales.

A New Physical Objection: How Much Energy Would a Simulation Require?

In 2025 Franco Vazza attacked the problem from another direction. Using known relations between information, energy and the physical limits of computation, he estimated the resources required for several scenarios: simulation of the visible universe, simulation of Earth, and even a strongly reduced-resolution Earth simulation compatible with certain high-energy observational constraints. [12]

His conclusion is strong: if the “parent” universe obeyed the same physics as ours and attempted to directly simulate the required physical content at the resolutions considered, the energy and computing requirements would be astronomical or physically prohibitive in his scenarios. Vazza therefore argues that a “Matrix” scenario implemented by a civilization in a physically similar universe faces severe physical constraints. [12]

That is a more serious result than internet claims about “glitches,” but its scope is not unlimited.

The 2026 Reply: How Much of the World Must Be Computed?

In April 2026 Eliott Edge and Chad Ashton Brown published a commentary on Vazza’s paper. They did not dispute his numerical conclusions for the physical architectures he analyzed. Instead they challenged the target of the argument. Bostrom’s original simulation hypothesis does not necessarily require microphysical computation of every cubic centimeter of the observable universe at extreme resolution. What matters for his argument is that simulated people receive sufficiently coherent experiences without detectable anomalies. [13]

This raises the possibility of adaptive or selective rendering, in which details are generated only when observationally relevant. Whether such a system could genuinely reproduce a world with our physics and conscious beings remains unproven. But the commentary shows why Vazza’s resource calculations cannot by themselves eliminate every Bostrom-style simulation scenario. [13]

This provides a good example of disciplined reasoning: Vazza places strong constraints on particular simulation architectures; the 2026 commentary points out that constraining one architecture is not equivalent to a logical refutation of the entire philosophical hypothesis.

What About “Glitches,” Déjà Vu and the Mandela Effect?

Videos of strange coincidences, déjà vu, perceptual errors, unusual memories or the so-called Mandela effect are often presented as “glitches in the Matrix.” But an anomaly becomes scientific evidence only when there is a prior prediction, a measurable phenomenon, independent reproducibility and a way to distinguish the simulation explanation from ordinary psychological or physical alternatives.

At present, anecdotal events of this kind are not evidence for simulation. A hypothesis that can explain every anomaly after the fact by saying “the simulator did it” loses predictive power. If a simulator can arbitrarily alter memories, measurements and laws, the hypothesis becomes almost indistinguishable from a metaphysical skeptical scenario from inside the system.

Perfect Simulation and the Problem of Testability

This is the central epistemological problem. If the simulation is perfect and its creator does not want to be detected, every experiment we perform could simply be part of the simulated world. No internal result would necessarily provide access to the “outside” level. Such a possibility can be philosophically coherent while being empirically weak because it produces no distinguishing prediction.

A narrower hypothesis — for example, “spacetime uses a cubic lattice at a measurable scale” — becomes scientifically interesting precisely because it can fail. The paradox is that the most comprehensive version of the simulation hypothesis is the hardest to test, while the most testable versions are far narrower than the general statement “we live in a simulation.” [4][12][13]

Would a Simulated World Be Unreal?

Even if we someday discovered that our world was simulated, it would not necessarily follow that it was “unreal.” For inhabitants of the simulation, stones would still injure, relationships would still have consequences, pain would still be pain, and decisions would still shape the future. In contemporary philosophy David Chalmers has argued that virtual worlds are not automatically false or valueless; a computationally realized reality could still be real at a different ontological level. [14]

The simulation hypothesis is therefore not necessarily nihilistic. If true, it would primarily relocate the question: what does our world depend on ontologically, and what is the relationship between our level and any external one?

What Can We Actually Say Today?

Bostrom’s argument is a philosophical trilemma, not experimental evidence. Its force depends on strong assumptions about consciousness, future civilizations, computing capacity, motivation and the statistical sampling of observers. Bostrom himself did not claim to have proved that we are simulated. [1][2]

Digital physics, informational physics and quantum computation are not the same as the simulation hypothesis. Zuse, Fredkin, Wheeler, Deutsch, Lloyd and ’t Hooft have explored the relationship between information, computation and physical law in very different ways. Even if one of the strongest computational ontologies turned out to be correct, an external simulator would not follow automatically. [5][6][7][8][10][11]

Physics can constrain some specific simulation scenarios. Lattice models can predict symmetry-breaking signatures, and physical limits on computation create severe difficulties for naive scenarios in which a civilization inside a universe like ours simply calculates every element of our universe at full resolution. [4][12]

But the general, perfectly adaptive simulation hypothesis currently has no confirmed empirical test. If formulated so that a simulator can hide every trace and generate every possible result, it becomes closer to a metaphysical framework than an ordinary physical theory.

The most important boundary is therefore simple: “the universe can be described in terms of information” ≠ “the universe is a computer program” ≠ “that program is being run by an external simulator.” The first proposition has deep connections with modern physics. The second is a serious but unproven ontological possibility. The third remains an intriguing philosophical hypothesis without direct evidence that our world actually works that way.