Skip to content
Above Top Secret
Science & The Unexplained

The simulation argument is not the claim you think it is

Nick Bostrom's paper is quoted everywhere as saying we probably live in a simulation. It says something narrower and more interesting, and the difference is the whole argument.

ATS Research DeskDocuments and archives
3 min read
A data centre

Bostrom's 2003 paper argues for a disjunction, not for a conclusion.

The philosopher Nick Bostrom published the simulation argument in 2003. It is one of the most cited and most misreported philosophy papers of the century.

What it actually argues

A trilemma. At least one of the following three propositions is true:

Illustration

One: almost all civilisations at our stage go extinct before reaching the technological capability to run detailed ancestor simulations.

Two: almost all civilisations that reach that capability choose not to run such simulations in significant numbers.

Three: we are almost certainly living in a simulation.

Why the structure matters

Because each branch is interesting.

If proposition one holds, there is something that reliably kills technological civilisations, which bears directly on the Fermi paradox and on existential risk.

If proposition two holds, advanced civilisations converge on declining to do something they are capable of — an ethical prohibition, or a loss of interest, that holds almost universally. That is a strong claim about the trajectory of intelligence.

The argument is valuable because it forces a choice between three uncomfortable options, not because it delivers one.

Can it be tested

Attempts have been made and they are more limited than usually presented.

One proposal, from Beane, Davoudi and Savage in 2012, notes that if a simulation used a discrete spacetime lattice — as lattice quantum chromodynamics calculations do — it might impose a signature such as an anisotropy or a cutoff in the spectrum of high-energy cosmic rays.

The limitation is severe: it tests one specific implementation. A simulation not using a lattice, or using one far finer than we can probe, produces no such signature. A negative result rules out a method, not the hypothesis.

The substrate-independence assumption

The argument depends on a premise that is doing a great deal of unnoticed work.

It assumes that consciousness arises from the pattern of computation rather than from the physical material performing it — so that a sufficiently detailed simulation of a mind would be a mind, with experiences.

If that is false, simulated people are not observers, the population of observers does not include them, and the counting argument collapses entirely.

What it shares with older ideas

A great deal. Descartes's evil demon, Zhuangzi's butterfly dream, and the brain in a vat are the same structural problem: no observation made from inside a deceptive system can establish that the system is not deceptive.

Bostrom's contribution is to give it a probabilistic form tied to a specific technological capability, which makes it discussable in a way the older versions were not.

Join the discussion

Still open on this one:

  1. 01

    Does the argument survive without the indifference assumption?

    It relies on treating oneself as a random sample from the set of observers with our experiences. That step is contested in the philosophical literature and is where the technical argument actually happens.

Join the discussion

Be the first to comment

Sources

Open them and check for yourself. Where a record is public, it is linked.

  1. NASA science

    NASA

0 comments