Ancestor Simulations Theory contradicts Chaos Theory, Quantum Mechanics and Irreducibility?

Question

(NB. I've no training as a philosopher, but I'm a student of science with an interest)

So there in the media this theory of reality as a simulation is gaining popularity, mostly because of types as Elon Musk supporting the theory. This video gives a nice overview of the argument as I mostly encounter it:

https://www.youtube.com/watch?v=JB3v3bDba5g

However, it seems to me that the argument is ridiculous. The idea is that we develop simulations that somehow "completely" capture our reality, or are at least qualitatively indistinguishable, and that we would use these simulations to map out the history of our species or something like that. Now it seems to me such ancestor simulations contradict everything we have learned about the physical world and dynamical system theory, including the inherent uncertainty of quantum mechanics, the problem of divergent paths from chaotic theory and the issue of irreducibility as laid out by for example Steven Wolfram.

Now I imagine that even when taking the ancestor story with a grain of salt, we may want to simulate human brains to study certain interactions or to create a game or something. But these simulations would always be (much) simpler than the reality in which they are constructed, and so the notion of simulations in simulations in simulations becomes complete nonsense.

I realize that you can circumvent this by assuming that the reality within which our reality is simulated is much more complicated than our own. But this element seems to be completely absent from the popular discourse, and it also makes the argument much less convincing (the whole point is that there is a very long sequence of simulations, this is necessary to make a statement like "the odds are very high that we live in a simulation").

So why do all these apparently briljant people support this theory? Am I missing something? Or are they missing something? I realize that my counterarguments are far from original, so Elon Musk must have been confronted with these, right?

Answer 1

I've been studying this particular theory recently. While I agree that it has significant flaws at a deep level, it's harder to debunk than it might initially seem. You might want to read the originating paper, by philosopher Nick Bostrom, it is neither long nor difficult.

In the case of your particular objection, the answer is that the simulation proposed is not an exact, full simulation of an entire universe, but rather the simulation of a portion of the universe, with the rest faked up. The obvious analog is a video game. Usually you're supposed to feel as though your video game character exists in a big huge world, but all that is actually simulated is the portion you interact with. So a person in a simulation could potentially actually be in a smaller, simpler universe than the one he thinks he inhabits.

This doesn't account for the concept of indefinitely nested simulations, however, and in fact, the core theory does not depend on nested simulations. The very large number of people in simulations is a result of the simulators running many simulations, not because each simulation is necessarily part of a large chain. Bostrom does consider the possibility of nested simulations (explaining it largely with some vague handwaving), but only as a non-functional coda to the main argument. That hasn't stopped it, of course, from capturing people's imaginations!

Answer 2

The simulation argument is false, but your statement about why it is false is wrong. Your argument is the following:

However, it seems to me that the argument is ridiculous. The idea is that we develop simulations that somehow "completely" capture our reality, or are at least qualitatively indistinguishable, and that we would use these simulations to map out the history of our species or something like that. Now it seems to me such ancestor simulations contradict everything we have learned about the physical world and dynamical system theory, including the inherent uncertainty of quantum mechanics, the problem of divergent paths from chaotic theory and the issue of irreducibility as laid out by for example Steven Wolfram.

Quantum mechanics (QM) is not about uncertainty. The reason commonly given for saying QM is about uncertainty is that in most experiments there is more than one possible outcome and there is no way to predict which outcome you will see. In QM muultiple versions of the same system can exist at the same time. For example, in a single particle double slit interference there is a version of the particle going through each slit. These different versions of the same particle were initially in the same state and could not be distinguished. In QM each particle exists in multiple instances all the time, sometimes those instances are distinguishable, sometimes they are not. QM predicts how the instances of a particle will be partitioned up into different versions over time and assigns a numerical quantity to each version of a particle that roughly measures how much there is of one version compared to another. The numbers saying how much there is of a given version obey the rules of probability and can be used to make decision theoretic predictions about what a rational agent should do. See

https://conjecturesandrefutations.com/2015/04/04/fungibility-in-quantum-mechanics/

"The Beginning of Infinity" by David Deutsch, Chapter 11.

"The Fabric of Reality" by David Deutsch, Chapter 2.

https://arxiv.org/abs/1508.02048.

Quantum systems can be simulated to any desired degree of accuracy by a universal quantum computer:

http://www.daviddeutsch.org.uk/wp-content/ItFromQubit.pdf.

And since the world is governed on all scales by quantum mechanics, false claims to the contrary notwithstanding, all physical systems can be simulated by a universal quantum computer.

You say that it is ridiculous to suppose that a human brain can be simulated but this is not true. We don't know how to do such a simulation but the laws of physics are incompatible with the idea that it can't be simulated. Also, since computers are physical systems, one computer can simulate another and there is nested simulations are possible.

But the simulation argument is false for two reasons.

First, a universal computer can be made out of any hardware capable of performing a certain finite set of operations. It follows that if the simulation argument is true, we can never discover anything about the hardware the simulation is running on. As such, the simulation argument can't solve any problems in philosophy or physics or any other subject and it implies that the real world is incomprehensible since we can't understand how the world, i.e. - the simulator, actually works. The simulation argument is as useless and irrational as the bare assertion 'god did it.'

Second, the alleged probability of being in a simulation is pure fluff. The authors of the argument provide no explanation of how to get numbers from the simulation argument to provide such probabilities. This is a serious problem that destroys their ability to explain or predict anything. Consider the computations taking place on my computer right now. These computations may be represented redundantly by many electrons flowing round various circuits. So how do we count the number of computations? Does my computer count as one computation? But the information and the computation is represented redundantly, so why not count it as several computations?

These problems have been pointed out several times by David Deutsch, e.g. - "The Beginning of Infinity" starting around p. 467, and I haven't seen any reply from advocates of the simulation argument.