QuestionsIs the world real?Computing all possible universes is cheaper than computing just one
Is the world real?Probably a simulation

Computing all possible universes is cheaper than computing just one

The Great Programmer; digital physics; the universe is computed (Zuse, Schmidhuber)

Jürgen Schmidhuber born 1963Computer scientist: our universe may be computed by a very short program, which also computes every other possible universe.
See profile
What it claims

It is reasonable to think our universe is computed by a short program, among all computable universes: we are, in this sense, in a simulation.

Why

A short program that runs every possible universe needs less information than describing just one of them: it is simpler.

The same idea through history
Konrad Zuse1969Zuse is the first to propose, in 1967 (book "Rechnender Raum" in 1969), that the history of the universe is computed by a computer, perhaps a cellular automaton. We did not read him directly: this is Schmidhuber's summary.
Edward Fredkin1985Fredkin takes up the idea in the 1980s ("digital physics"), according to Schmidhuber's page; we did not read his texts.
Jürgen Schmidhuber1997In 1997 Schmidhuber posits a Great Programmer running all computable universes, and adds an argument from simplicity: describing all of them takes less information than describing one. In 2000 he defends the hypothesis itself: it is the simplest explanation compatible with what we know.
Max Tegmark2008Tegmark pushes the idea further: the computer does not even need to compute, it just has to describe the mathematical structure of the universe.
In short

"Computing all possible universes is cheaper than computing just one" is an argument by Jürgen Schmidhuber (Schmidhuber, 1997 and 2000, 1997) on the question "Is the world real?" (answer: probably a simulation). Its claim: It is reasonable to think our universe is computed by a short program, among all computable universes: we are, in this sense, in a simulation. According to AskPhilosophy's logic check (Z3 solver), the argument is valid only if one grants a premise the author leaves unstated (P4). AskPhilosophy presents 2 objections, 1 of them with a documented reply, and the original text behind each premise, checked word for word.

Schmidhuber, 1997 and 2000
Valid with an implicit premise

In 1997 the computer scientist Jürgen Schmidhuber imagines a "Great Programmer" who, long ago, wrote a program that runs all possible universes, meaning all computable ones, on his big computer. Ours would be one of them. His argument is one of simplicity. Most possible universes are pure disorder, and it would take a very long program to describe even one. But the program that runs all of them, one after another, is very short. In 2000 he adds the missing piece: nothing we know forces us to think the universe is incomputable, and a theory in which it is computed is the simplest of all. So one may believe it until evidence comes against it. A caveat: the Great Programmer is an image; the argument is about the length of programs, not about who would run them. This is "probably a simulation" in the sense of "a computed world", not in Bostrom's sense (descendants simulating us).

P1Nothing we know forces us to think the universe is incomputable: a history produced step by step by a program is the simplest explanation compatible with what we observe.
P2Almost all computable universes are pure disorder: describing one particular universe takes a very long program. By contrast, the program that runs them all is very short.
P3When two explanations exist, the simpler is better: a set-up where all possible universes are computed is simpler than one where only ours is.
P4 · implicit premiseIf the best explanation of our universe is that it is the output of a program, then we live in a computed, that is simulated, world.
So it is reasonable to think our universe is computed by a short program, among all computable universes: we are, in this sense, in a simulation.

click a premise to reject it

If you accept every premise, logic forces you to accept the conclusion. So the debate is about the premises.
Checked by a logic solver: the premises cannot all be true while the conclusion is false.
Without the implicit premise: Without P4, nothing prevents the simplest explanation of the universe from being a program without making us "simulated" beings: a program may merely describe the laws, with nobody running it.
P1: nothing forces us to think the universe incomputable, and thinking it computed is simplest. P2: computing all universes costs less information than computing one. P3: simplicity decides between explanations. P4, hidden, says a universe whose best explanation is a program is indeed a computed world. So it is reasonable to believe we are in a computed world. What is debated: P1 (is physics computable?) and the gap between "computed" and "simulated by someone".
Analogy. Think of a library holding every possible book. It can be described in one sentence ("all strings of letters"), while describing a single, random-looking book means copying it whole. Schmidhuber's idea is the same: computing everything is simpler than picking one universe at random. (Our image.)

Objections

2

Each objection is a conversation: click to see who replies and where the debate stands.

Roger Penrose 1989 · attacks premise 1

Penrose argues that conscious thought is not reducible to computation, and that a new, non-computable physics will be needed to understand it. If that is true, the universe is not the output of a program, and P1 falls.

Schmidhuber replies
Schmidhuber
Schmidhuber answers that no evidence forces incomputable quantities to play a role in explaining the world. One may believe the universe computable until someone shows it could not exist without incomputable numbers.
The debate goes on (our summary) The answer is general: Schmidhuber does not discuss Penrose's own argument, which starts from mathematical understanding and Gödel's theorem, not from a physical observation (our summary).
Where the debate stands (our summary)Open point: we do not know whether the laws of physics are entirely computable. Schmidhuber puts the burden of proof on Penrose; Penrose thinks he has met it with Gödel, which most logicians dispute (our summary).
S
Sabine Hossenfelder 2021 · attacks premise 1

Hossenfelder grants that "our universe computes its laws" is nothing shocking, but notes that nobody today knows how to recover general relativity and the Standard Model with a step-by-step algorithm. Such attempts are generally incompatible with Einstein's symmetries.

no reply
We have not yet found a direct reply by Schmidhuber in the texts.
The debate goes on (our summary) Hossenfelder writes in 2021, after Schmidhuber (1997); to our knowledge he has not answered this text. She mostly targets the idea of a programmer running a simulation, more than Schmidhuber's cautious thesis (our summary).
Where the debate stands (our summary)Underlying disagreement: "the universe obeys mathematical laws" is widely accepted; "it is the output of an algorithm reproducing known physics" has not been shown. Open point (our summary).