Are You Living in a Computer Simulation?
The phrase simulation theory has come to mean almost any suspicion that reality is not what it appears to be.
Maybe the universe is a program. Maybe consciousness has been placed inside an artificial world. Maybe reality has layers, and ours exists inside something larger. These are different hypotheses. Giving them one name does not give them the same evidence.
Nick Bostrom’s Simulation Argument1 is narrower. It does not begin with glitches in the Matrix, quantum mechanics, or the intuition that the universe looks programmed. It begins with a possible future for civilizations like ours and asks a question about populations:
If advanced civilizations can create conscious simulations of their ancestors, and if they create enormous numbers of them, why should we expect to belong to the small original population rather than one of the simulations?
That is the argument. The claim that we are simulated is only one possible answer.
The argument is not the hypothesis
Bostrom’s conclusion is a trilemma. At least one of the following must be true:
- Almost no civilization at our stage survives to become technologically mature.
- Almost no technologically mature civilization creates a significant number of ancestor simulations.
- Almost all observers with experiences like ours are simulated.
The possibilities can overlap, and the argument does not assign each of them a one-in-three probability. It says we cannot confidently deny all three at once.
Suppose civilizations like ours usually survive long enough to acquire enormous computational resources. That makes the first proposition unlikely. Suppose at least some of the survivors use a small portion of those resources to simulate conscious people living through histories like their own. That makes the second unlikely.
The simulations would then multiply. There may be only one original history, but a mature civilization could run millions of variations, each containing billions of conscious observers. Simulated people with human-type experiences could vastly outnumber the biological people who lived before the technology existed.
If we have no observation that distinguishes us from those simulated observers, Bostrom argues that our belief should track the population. If almost everyone with experiences like ours is simulated, we should expect to be simulated too.
The conclusion is not produced by increasingly realistic video games. It is produced by multiplying conscious observers.
What has to be true
The counting works only if several substantial premises survive.
A simulated person must be conscious
A weather simulation does not get wet. A simulated fire does not burn the computer running it. Why should a simulated brain produce an experience rather than merely calculate what a conscious person would do?
Bostrom assumes substrate independence: reproduce the relevant structures and processes of a human brain at sufficient detail, and conscious experience can occur in that system. The mind need not be made of biological cells.
This is not a conclusion of the Simulation Argument. It is imported from the philosophy of mind. If a detailed ancestor simulation contains convincing characters but no conscious observers, it adds nobody to the population. The counting argument collapses.
Simulations must be affordable
Bostrom uses posthuman to describe a civilization with capabilities far beyond ours. Such a civilization might convert planetary or astronomical resources into computation. It would not need to simulate every particle in the universe, only the minds of its inhabitants and enough of their environment to make their experiences coherent.2
The argument depends on a vast gap between the computation available to a mature civilization and the cost of reproducing one human history. If conscious minds cannot be cheaply reproduced, environments are much more demanding than expected, or physical limits prevent civilizations from accumulating the necessary machinery, the trilemma loses its pressure.
Someone must choose to run them
Capability is not motive.
Advanced civilizations might consider conscious simulations unethical. They might find them useless, prohibit them, or lose every desire we can imagine for creating them.
But abstention must be nearly universal. If a tiny minority can create enormous simulated populations, the exceptions may still outnumber all original observers. Bostrom does not need everyone to want an ancestor simulation. He needs a few civilizations to want many.
This makes speculation about posthuman motives unavoidable and unreliable. Research, entertainment, nostalgia, and education are motives available to us. A mature civilization may no longer contain anything psychologically continuous with present-day humanity.
We must know who counts as someone like us
Even if simulated minds are conscious and numerous, why should their existence tell me what I am?
Bostrom applies a restrained principle of indifference. If some fraction of observers with human-type experiences are simulated, and my experience gives me no reason to think I belong to one group rather than the other, my confidence that I am simulated should track that fraction.
The difficult phrase is “observers with human-type experiences.”
Do we count only historical replicas? Every biological human? Digital minds with altered memories? Brief conscious moments? Beings with unfamiliar senses? A vast population of minds unlike us does not obviously make us more likely to be simulated.
This reference-class problem is why Bostrom focuses on ancestor simulations. Broaden the category, and simulation becomes easier to imagine but harder to calculate.
What the argument cannot tell us
Bostrom gives us a constraint, not a detector. He does not identify which branch is true, establish that computation can create consciousness, or tell us what advanced civilizations will value.
He also does not give us a way to estimate every possible creator.
We cannot extrapolate from humanity’s computers and assume that a being outside our reality followed the same technological path. If our world is artificial, its laws may describe the constructed environment rather than the machinery beneath it. Its creators could face different resources, limits, and motives—or be so unlike us that those words no longer transfer.
Bostrom avoids that problem by making a narrower argument about civilizations with histories and observers like ours. If we are not simulated, our physics can help us reason about what our descendants might build. If we are simulated, the third branch is already true. That protects the trilemma, but it does not produce a probability for arbitrary parent realities.
The Fermi Paradox has a similarly limited role. An apparently unclaimed galaxy may tell us something about how often technological civilizations survive or become conspicuous. That bears on the first branch. It tells us nothing directly about creators outside our observable universe.
What if we are not the ancestors?
The usual simulation story imagines future descendants reconstructing beings like themselves. But perhaps the inhabitants are not reconstructions at all.
Perhaps the minds are the product.
Perhaps the world exists because artificial minds need somewhere to develop.
That is no longer Bostrom’s ancestor-simulation argument. His population estimates cannot simply be carried over: we do not know how many such minds would be created, whether they belong in our reference class, or why their creators would make them.
But his argument exposes the questions we need to answer. Can an artificial mind be conscious? Would it need an experienced world? Why would anyone create it? And what, if anything, can a mind inside that world learn about its origin?
Bostrom’s argument does not choose a door for us. Its value is that it shows us where the hinges are.
Notes
-
Nick Bostrom, “Are You Living in a Computer Simulation?”, The Philosophical Quarterly 53, no. 211 (2003), pp. 243–255. ↩
-
Nick Bostrom, “The Simulation Argument FAQ” (2025), especially the distinction between the Simulation Argument and the Simulation Hypothesis and the discussion of computational feasibility. ↩
Filtered post view · schema v1
Argument map
Loading the claims, evidence, objections, and assumptions behind this post…
- Supports / entails
- Attacks / contradicts
- Structural relation
- Question
- Claim
- Hypothesis
- Evidence
- Objection
- Assumption
- Conclusion
Drag to pan · scroll or pinch to zoom · select a node for details