The argument is not a probability pulled from a computer
The simulation hypothesis is easy to state and difficult to test: perhaps everything we observe is produced by machinery outside our universe. That possibility is older than computers in spirit. Dreams, Descartes’s deceiver, and the brain in a vat all ask how experience could reveal whether its apparent cause is genuine.
Nick Bostrom’s 2003 argument adds a population calculation. It does not begin with suspicious pixels, quantum mechanics, or a feeling that history has become implausible. It begins by imagining civilizations that survive long enough to gain extraordinary computing power and then use some of it to run detailed histories populated by conscious minds.
If one original civilization creates billions of such worlds, simulated observers would vastly outnumber observers in the original world. A person who knew only that they were one human-like observer might then assign a high probability to being among the larger group.
That “if” carries most of the weight.
Bostrom’s conclusion has three branches
The argument says at least one of three propositions must be close to true. First, almost no civilization at our stage survives or develops into a technologically mature civilization. Second, civilizations that do mature run few or no ancestor simulations. Third, simulated observers become so numerous that an observer like us is probably simulated.
This structure is often flattened into “Bostrom proved we live in a simulation.” He did not. The first two branches are not decorative escape clauses. A civilization could destroy itself, abandon growth, face physical limits, prohibit worlds containing suffering, or simply prefer other projects. Any of those possibilities reduces the number of simulated observers.
The third branch also requires more than convincing game graphics. The simulated people must actually have experiences. Characters that say “I am conscious” do not affect the probability calculation if nobody inside feels anything.
Computation does not settle consciousness
Bostrom assumes a form of substrate independence: reproduce the relevant organization of a mind closely enough and consciousness can occur in silicon or another medium. That position is serious, but unresolved.
Our article on the hard problem of consciousness explains the missing step. A computation may reproduce discrimination, memory, speech, and behavior. Philosophers still disagree over whether matching those functions guarantees an inner point of view. If biological properties are essential, an ancestor simulation could contain persuasive human behavior without multiplying conscious observers.
Counting creates a second problem. Should the reference group include all human-like minds, every conscious being, exact copies of you, or short-lived minds produced during failed runs? Different choices change the sample from which “you” are supposedly drawn. The arithmetic becomes precise only after a philosophical decision about who belongs in the total.
Physical limits constrain scenarios, not the whole idea
A computer inside a universe like ours cannot casually store and update every detail of an equally large universe. Research on physical resource requirements makes full-resolution copies look enormously expensive. That matters against a simple picture in which our entire observable universe is calculated particle by particle on familiar hardware.
It does not eliminate every version. A simulation might calculate only observable regions in detail, use approximations, or run under physical laws unlike ours. Each adjustment makes the claim harder to exclude—and less specific.
This is why “the universe has a speed limit” is weak evidence. The speed of light may resemble a bandwidth cap, and quantum outcomes may resemble detail filled in when observed. Resemblance is not a prediction. Physical laws look rule-bound because laws describe regularities; no external computer follows from that fact.
Researchers can instead define a particular mechanism. One proposal asked whether a universe calculated on a cubic lattice might leave directional patterns in extremely energetic cosmic rays. A matching pattern would require ordinary physical explanations to be ruled out. A missing pattern would reject that lattice model, not every possible simulator.
A perfect simulation inherits the old skeptical problem
Suppose the simulation reproduces every observation that base reality would produce. An experiment finding smooth space, ordinary atoms, and no error messages is then exactly what the simulator generates. No internal observation can distinguish the two worlds.
Such a hypothesis can remain logically possible while contributing little as science. A scientific model gains value by risking failure: it predicts some results and excludes others. An unspecified simulator able to produce any result excludes nothing.
This does not mean skepticism has been defeated. It means possibility and evidence must stay separate. We cannot prove that a perfect deceiver is absent, but that inability is not positive evidence that one is present.
Simulated does not have to mean unreal
David Chalmers argues that a simulated table could still be a real table within its world. It would be made from a different underlying structure than residents assume, but it could support real cups, meals, accidents, and conversations. Discovering that water consists of molecules changed what water is made of; it did not show that nobody had ever been wet.
The same point matters morally. If simulated beings are conscious, their pain does not become harmless because it is implemented differently. Ending their world could be killing, not closing an empty program. Creators outside the universe would not automatically be wise or good, and their existence would only move the question “why is there anything?” to another level.
So are we simulated? It is a coherent possibility supported by a conditional argument, not by a confirmed physical trace. Until a specific model makes a successful distinctive prediction, the honest answer remains open. Whatever its substrate, this is the world in which our choices currently reach other minds.
