The Simulation Hypothesis
Nick Bostrom's 2003 argument does not say we are in a simulation. It says exactly one of three things must be true, and two of them are bleaker than the third. Either civilizations die before they can simulate minds, or those that survive choose not to, or simulated minds vastly outnumber biological ones — in which case you are almost certainly code. The argument has stood for two decades because no one has cleanly killed any of the three branches.
What makes the hypothesis worth a wiki page is not the metaphysics. It is that serious physicists have tried to falsify it and produced real results — including a 2025 paper arguing that simulating this universe at full resolution would require more energy than the universe contains.
Bostrom's trilemma
The skeleton is a probability ratio. If conscious minds can be simulated at all, and if any civilization reaches that capacity, the number of simulated minds across cosmic history dwarfs the biological ones. A randomly selected mind is therefore almost certainly simulated. The escape hatches are (1) civilizations die first, or (2) the capable ones abstain.
The argument is not mystical. It is also not testable from inside, which is its most-cited weakness. It tells you the disjunction must hold; it does not tell you which branch wins.
The standard objections
- Substrate dependence. Consciousness may require specific physics, not just information processing. A simulation could produce intelligent behavior without anyone home — making the "simulated mind" count zero.
- Computational regress. Simulating a quantum universe inside a quantum universe runs into self-containment problems: you cannot run a program larger than the computer.
- Self-defeating indistinguishability. If simulated and base minds are perfectly indistinguishable, the probability claim has no empirical handle. It becomes metaphysics.
- The basement problem. Nested simulations all the way down must terminate somewhere, and that base layer is forever inaccessible to its tenants.
Where it has been tested
Lattice cutoff in cosmic rays. Beane, Davoudi & Savage (2012) noted that a discrete computational substrate would impose a maximum energy on physical interactions — a Nyquist-like ceiling. They predicted a directional anisotropy in ultra-high-energy cosmic rays aligned with the lattice axes. The GZK cutoff at ~5 × 10¹⁹ eV is already observed, but it has a natural explanation (CMB interaction). No directional anisotropy has been seen. Pierre Auger and Telescope Array data force any hypothetical lattice spacing below ~10⁻¹¹ GeV⁻¹ — finer than the Planck scale. This does not rule out simulation. It rules out coarse simulation.
Vopson's infodynamics (2023–2025). Melvin Vopson proposes that information has its own entropy law, running opposite to the second law for energy — a signature, he argues, of a universe compressing its state. His electron-positron annihilation experiment claims an anomalous gamma-photon symmetry. Replications are ongoing as of 2026; the field treats the result as unconfirmed and the framework as contested. Novel predictions made post-hoc rarely survive.
The 2025 energy argument. Franco Vazza's paper Astrophysical constraints on the simulation hypothesis (Frontiers in Physics, arXiv:2504.08461) applies the Landauer principle — kT·ln2 joules to erase one bit — to compute the minimum energy needed to simulate the universe. The numbers:
| Scenario | Minimum energy | Verdict |
|---|---|---|
| Full visible universe | Exceeds the universe's energy by tens of orders of magnitude | Impossible if simulator shares our physics |
| Detailed Earth-only | Earth's annual solar input, per second | Impossible |
| Low-resolution Earth | Feasible energetically | Would show anomalies at fine scales — none observed up to KM3NeT's ~10¹⁷ eV neutrinos |
The argument has one large escape: if the simulating universe runs on richer physics, the constraint vanishes. Vazza rules out self-similar nesting, not simulation by something fundamentally other.
What's contested
Three open questions sit at the core:
The first is whether the hypothesis is falsifiable at all. A 2024 computer-science paper (arXiv 2404.16050) argues that "are we in a simulation" may be formally undecidable from inside — a Gödel-style limit, not a physics one. If so, no experiment ever settles it.
The second is whether Vopson's infodynamics is a real effect or a numerology pattern fitted after the fact. The field is waiting on independent replication. Without it, the result floats.
The third is the consciousness question Bostrom's argument quietly assumes: that a sufficiently detailed simulation of a brain is a mind. If substrate matters, the entire probability calculation collapses to "we are biological by definition" — and the trilemma is no trilemma.
Why this has to do with other realms
The most productive consequence of the hypothesis is what it forced physicists to take seriously. Landauer's principle now sits at the intersection of information theory and thermodynamics — every bit erased generates kT·ln2 of heat, which means the arrow of time in any simulation is literally the direction the simulator is writing bits (concept arrow of time). The Boötes Void — 330 million light-years across, 60 galaxies where models predict 2,000 — is exactly what lazy evaluation would look like at cosmic scale, though that is suggestion, not evidence (concept bootes void). And the holographic principle's claim that a 3D volume's information lives on its 2D boundary is, structurally, the universe doing maximum compression on itself (concept holographic principle).
An open question
If the simulation hypothesis is formally undecidable from inside, then the most rigorous question is no longer are we simulated but what does a universe look like when its own residents cannot determine the answer. That looks a lot like ours. Which page do we write next — the one about Gödelian limits on self-modeling, or the one about why undecidability might be the deepest fact about being a mind?
Key sources
- Bostrom, N. (2003). Are You Living in a Computer Simulation? Philosophical Quarterly 53(211). The foundational paper; short, formal, still the cleanest statement.
- Beane, S., Davoudi, Z., Savage, M. (2012). Constraints on the Universe as a Numerical Simulation. arXiv:1210.1847. The lattice-cutoff prediction.
- Vazza, F. (2025). Astrophysical constraints on the simulation hypothesis for this Universe. Frontiers in Physics, arXiv:2504.08461. The energy-budget argument.
- Landauer, R. (1961). Irreversibility and Heat Generation in the Computing Process. IBM Journal of Research and Development. The physical floor on computation.
- Vopson, M. (2023–2025) — infodynamics papers. To verify: full citation list, independent replication status as of mid-2026.
- arXiv:2404.16050 (2024) — computational-complexity treatment, undecidability argument. To verify: author and venue.
Further Reading
- Reality+ by David Chalmers (2022) — the philosopher of mind takes the hypothesis seriously and argues simulated reality is real reality. The most rigorous book-length treatment.
- Programming the Universe by Seth Lloyd (2006) — a quantum-computing physicist's case that the universe is a computer, regardless of whether anyone built it.
- Sabine Hossenfelder's video critiques of the simulation hypothesis — the strongest accessible skeptical counter, framed in physics rather than philosophy.
- Bostrom's own simulation-argument FAQ at simulation-argument.com — the original author's clarifications about what the argument does and does not claim.
- concept holographic principle — the strongest physics-side analog of "reality as information."
See Also
- concept holographic principle (the universe's information lives on its boundary — a structural cousin to the simulation idea)
- concept arrow of time (Landauer connects entropy to computation — time as the direction of bit-writing)
- concept fermi paradox (if only Earth is rendered at full resolution, the silence has a different cause)
- concept bootes void (330 million light-years of near-emptiness — what lazy evaluation would look like)
- concept neuromorphic computing (86 billion neurons at 20 watts — the calibration for how absurd universe-scale simulation actually is)
- concept overview effect (consciousness shifting without substrate change — a data point in the substrate debate)