Review: Are You Living in a Computer Simulation?
This article reviews philosopher Nick Bostrom’s influential 2003 paper,
“Are You Living in a Computer Simulation?”
1. Almost all civilizations go extinct before developing the technology to create simulations.
2. Advanced civilizations capable of simulations choose not to create them.
3. We are almost certainly living in a computer simulation.
Bostrom’s simulation argument is a logical trilemma, meaning at least one of three possibilities must be true. Imagine you’re playing a video game like The Sims, where characters live in a virtual world. Now, scale that up to a hyper-advanced simulation where the characters (us) are fully conscious. Bostrom suggests:
The argument’s power lies in its probabilistic logic: if simulations are common, the odds of being in the one true reality are tiny, much like picking the single “real” server out of millions in a massive online game.
Bostrom frames the simulation hypothesis using probability. Let \( N_{\text{base}} \) represent the number of conscious beings in the base (non-simulated) reality, and \( N_{\text{sim}} \) the number of conscious beings in all simulated realities. The total number of observers is: \[ N_{\text{total}} = N_{\text{base}} + N_{\text{sim}} \] The probability that you are in a simulation is approximately: \[ P(\text{simulation}) \approx \frac{N_{\text{sim}}}{N_{\text{total}}} \] To illustrate, consider a hypothetical advanced civilization, “FutureTech,” capable of running simulations. Suppose FutureTech creates \( S = 10^6 \) simulations, each with \( B = 10^9 \) conscious beings (similar to Earth’s population). The total number of simulated beings is: \[ N_{\text{sim}} = S \cdot B = 10^6 \cdot 10^9 = 10^{15} \] If the base reality has \( N_{\text{base}} = 10^9 \) conscious beings, the total number of observers is: \[ N_{\text{total}} = 10^{15} + 10^9 \] The probability of being in a simulation becomes: \[ P(\text{simulation}) \approx \frac{10^{15}}{10^{15} + 10^9} \approx 0.999999999 \] This means there’s a 99.9999999% chance you’re in a simulation. Bostrom introduces the simulation indifference condition, defining \( f_{\text{sim}} \) as the fraction of observers in simulations: \[ f_{\text{sim}} = \frac{N_{\text{sim}}}{N_{\text{total}}} \] If \( f_{\text{sim}} \approx 1 \), we are almost certainly simulated. For example, if FutureTech runs just 1,000 simulations instead of a million, with \( S = 10^3 \), then \( N_{\text{sim}} = 10^3 \cdot 10^9 = 10^{12} \), and: \[ P(\text{simulation}) \approx \frac{10^{12}}{10^{12} + 10^9} \approx 0.999 \] Even with fewer simulations, the odds remain overwhelmingly in favor of being simulated.
In the main paper, Bostrom does not define variables like \( N_{\text{base}} \) or \( N_{\text{sim}} \). Instead, he frames the probability that we are in a simulation using the formula:
\[ f_{\text{sim}} = \frac{f_p f_1 N_1}{f_p f_1 N_1 + 1} \]
where:
This formula gives the estimated fraction of all observers with human-like experiences who are actually simulated. If the numerator is very large (i.e., many simulations), then \( f_{\text{sim}} \) approaches 1, meaning most such observers are in simulations.
We reframe this idea using an alternative but mathematically consistent model: by comparing the number of simulated observers to those in base reality using variables like \( N_{\text{sim}} \), \( N_{\text{base}} \), and \( P(\text{simulation}) \). This approach makes the argument more accessible through concrete numerical examples, but the philosophical conclusion remains the same.
Many civilizations may collapse before reaching a "posthuman" stage where they can simulate conscious minds. Think of historical empires that fell due to war or famine modern risks like nuclear conflict, climate change, pandemics, or rogue AI could similarly wipe us out before we develop simulation technology.
Even if civilizations achieve posthuman status, they might avoid running ancestor simulations. For example, they could view simulating conscious beings as unethical, akin to creating sentient beings who might suffer. Alternatively, legal bans, lack of interest in history, or prioritizing resources for other projects (like space exploration) could prevent simulations.
If advanced civilizations run ancestor simulations, the number of simulated minds would dwarf real ones. Imagine a supercomputer simulating billions of lives across thousands of virtual Earths, each person in those simulations feels real, just like us. The sheer volume of simulated beings makes it statistically probable that we’re in one of these virtual worlds.
Bostrom’s argument uses anthropic reasoning: if most beings with experiences like ours are simulated, we’re likely simulated too. This is similar to Descartes’ idea of an evil demon tricking us or the brain-in-a-vat scenario, but it’s rooted in modern ideas about computation and probability. For example, if 99.9% of all “humans” are in simulations, it’s rational to bet we’re one of them.
If we’re in a simulation, it challenges our understanding of reality, free will, and ethics. Are our choices truly ours, or scripted by simulators? Should simulators be responsible for our suffering, like programmers fixing bugs? If we’re not in a simulation, Bostrom’s argument warns of existential risks that could prevent us from reaching a posthuman stage or suggests advanced civilizations have surprising cultural limits. The idea also echoes religious notions of a created world, but with computers instead of gods.
The simulation argument has inspired movies like The Matrix, where humans live in a virtual reality, and Inception, with layered dream worlds. It fuels debates among philosophers and scientists, some of whom argue that we’re likely living in a simulation, about the nature of consciousness and the future of technology.
Bostrom’s argument doesn’t prove we’re in a simulation but shows that, unless civilizations collapse or avoid simulations, the odds favor it. It’s a mind-bending invitation to question consciousness, technology, and our place in the universe.
More information here simulation-argument.com.