YYoung and Profiting
← All frameworks
MindsetNick Bostrom

The Simulation Argument Trilemma

Test three exhaustive explanations for a simulation-rich future

Difficulty
Advanced
Time to result
~weeks to results
Steps
5
Confidence
99%

Begin with a future in which technologically mature civilizations can run detailed ancestor simulations containing conscious observers. If many such simulations are run, simulated observers with experiences like ours would vastly outnumber comparable biological originals. That leaves a trilemma: almost all civilizations go extinct before maturity; almost all mature civilizations refrain from running many ancestor simulations; or observers like us are almost certainly simulated. The argument does not select a branch by itself. Instead, evidence about extinction risk, technological progress, and future interest shifts probability among the three. The model therefore turns an unconstrained metaphysical question into a set of empirical and behavioral uncertainties that can be updated over time.

Origin

Nick Bostrom published the simulation argument in 2002 and explains its three possibilities in this Young and Profiting interview.

Core principles

  • 01Separate an argument's structure from the probability of each outcome
  • 02Treat observer counts as evidence when experiences are indistinguishable
  • 03Update each branch as new evidence arrives

How to run it

  1. 1

    Specify technological maturity

    Define the capability required to simulate environments and conscious minds at high granularity.

    Pro tip Use a concrete capability threshold rather than vague advanced technology.

  2. 2

    Test the extinction branch

    Ask whether nearly all civilizations at our stage fail before reaching that capability.

    Watch out One civilization surviving is not enough unless its simulation output is large.

  3. 3

    Test the restraint branch

    Ask whether nearly all mature civilizations would avoid devoting even a small share of computation to ancestor simulations.

    Pro tip Consider scientific, entertainment, cultural, historical, and moral motives.

  4. 4

    Compare observer populations

    If both filters fail, compare the number of simulated observers like us with original biological observers.

    Watch out The inference depends on simulated minds having experiences relevantly like ours.

  5. 5

    Update without overclaiming

    Shift probability as evidence weakens or strengthens the first two branches, while keeping the argument distinct from a precise probability estimate.

    Pro tip Track which branch each new observation bears on.

    Watch out Do not present the trilemma itself as proof that the simulation branch is true.

In the wild

Progress toward conscious simulations

Suppose civilization develops increasingly detailed virtual environments, can model brains, and continues to express strong interest in historical worlds. Those developments weaken both the technological-failure and universal-disinterest branches, shifting probability toward the simulation branch without proving it.

A directional probability update rather than a categorical conclusion.

Common mistakes

Forcing a precise probability

The trilemma establishes that at least one branch obtains; it does not supply all priors needed for an exact percentage.

Treating glitches as decisive evidence

Ordinary hallucination, memory error, misinterpretation, or fabrication can explain reports without invoking simulation defects.

Is it for you?

Best for

It is best for disciplined reasoning about futures that create many indistinguishable observers.

Not ideal for

It is not ideal for proving that any reported anomaly is a simulation glitch.

From the transcript

the simulation argument itself that you're referring to the paper that was published in 2002 um only demonstrates that one of three possibilities obtains one…

Nick Bostrom · 04:00

if most people with our kinds of experiences are simulated then we should think we are probably amongst the simulated ones uh rather than the…

Nick Bostrom · 05:00

But if both of those fail then we end up with a third possibility that [snorts] uh that we are almost certainly currently living in…

Nick Bostrom · 07:00

From the episode

Nick Bostrom: How Entrepreneurs Can Win in an AI-Dominated World

Nick Bostrom