Elijah Mirecki
Introduction
Once in a while, you hear someone propose “maybe our universe is in a (computer) simulation”. It goes something along the lines of “we can already basically simulate the universe, so it’s not a stretch to assume we’re in a simulation”.
That’s absurd. In this post, I’ll rebut the idea from several perspectives.
1. The Principle of Parsimony indicates we’re not actively running in a simulation
The Ruliad is defined as “the result of following all possible computational rules in all possible ways”. If our universe is computable (we can simulate it), then it must be defined in the Ruliad.
- S: We’re running in a simulation
- C: Our universe is computable
- R: Our universe is in the Ruliad
$$S \Rightarrow C$$ $$R \iff C$$ $$\therefore S \Rightarrow R$$
By Bayes Theorem:
$$(P(S) \land P(R)) \le P(R)$$
i.e., it’s more likely for us to be in the Ruliad than both in the Ruliad and in a simulation.
So by the Principle of Parsimony, if we assume it’s possible that we can be running in a simulation - i.e., if the universe is computable - then we should really ditch the idea alltogether and believe that we’re a purely mathematical construct.
2. Simulate the universe? Try simulating the 3-body problem first
We have no mathematics for simulating the general 3-body problem. We must rely on approximations for orbital mechanics simulations.
3. To simulate a star, you need a star
The maths behind simulating a SINGLE PAIR of hydrogen atoms takes a full page of partial differential equations to solve. Our current compute technology is pitifully slow at simulating Quantum Mechanics, so we’d likely need to use a quantum computer. If you’re extremely optimistic, maybe we could simulate a single particle using a single particle in our own universe. If we need a 1:1 mapping from our own universes particles to the nested simulation’s particles, then we would literally need a computer the size of a star to simulate a star, and that’s being optimistic. Realistically, your simulation supercomputer would have some overhead. Good luck creating a galaxy sized computer to simulate a single galaxy.
Now, what if the universe we’re being simulated in has different Physics? Then my friend, the initial premise that “we can basically simulate our own universe” is false.
4. Infinite recursion of inner simulations
By the title of this section you can probably already guess where it’s going. If we can simulate our universe, then inside that universe we should be able to self-simulate (assuming we can somehow figure out the universe’s starting conditions).
Suppose we can run the inner simulation faster than our own universe. In other words, suppose our universe in computationally reducable.
This scenario has the most problems. If we can simulate our universe faster than we experience, we know the universe is computationally reducable, meaning there are short cuts to jump forward in time. We have no evidence this is possible, but let’s assume it is.
So you create a nested simulation that runs at 2x speed of our current universe. What happens when the nested simulation hits the point in time where you created that simulation? Then you have a nested-nested simulation. The computational resources are double since you’re effectively running two now. And worse, imagine if multiple people in our universe run self-simulations simultaneously, then we’re going into exponentially expensive simulation territory!
Poof, our universe as we know it is destroyed, since the computer running our currently simulation must grind to a halt with the progression of simulating infinite amounts of nested simulations.
Now let’s talk Quantum Mechanics. The most widely accepted interpretation is that wavefunctions “collapse” into a value in their probability distribution. The only way to perfectly simulate our own universe in this interpretation would be if we had a different model of quantum mechanics that supports perfect predictability. So we’re stuck, and can only figure out the most likely future scenarios, rather than a perfect future prediction.
What about the many worlds interpretation? If that’s the REAL interpretation of Quantum Mechanics, and we know what world line we’re running on, then we can run a simulation of the same world line. Now suppose we can measure what worldline we’re in, somehow. In that case, we actually could create a perfect nested simulation.
The many worlds interpretation doesn’t make sense in a simulation. You would need fractally expanding compute power at every planck time length.
Tangentially, the Ruliad would easily support the many worlds interpretation, and I’ve become more fond of it after reading some essays by Eliezer Yudkowsky.
5. Can we commit to a billion year project?
Now suppose we run the nested simulation at the speed of our universe or slower. In this case, we have no predictive power since we can’t advance the simulation faster than reality advances. Booooring!
Suppose our overlords have a super efficient computer and manage to run our simulation at exactly the same speed as their own universe. They would need to run their simulation for BILLIONS of years for us to exist. Humans are extremely recent in universe evolution timescales.
… we marvel at ancient wonders that took hundreds of years to complete. Do you really expect a civilization to commit to running a simulation for billions of years?