Do we live in a simulation? Probably not.

A while back I came across the article: The Lowest-Bid Universe, where the idea of whether we live in a simulation is examined. The premise of the text is that if this really is the case, then we should lodge complaints of our own reality. Because, apparently, the reality we live in is not that well made.

The article starts by superficially mentioning some of the wilder ideas in physics, and then moves on to a "more philosophically motivated" idea that the reality as we experience it is the product of an experiment or a simulation. The standard motivation behind this is to argue that even if only a few civilizations evolve so far that they can produce simulations of whole universes, they inevitably will do a large number of them. Therefore, if there are a large number of simulated realities, it is very likely that our reality is also simulated.

Then the writer concludes that if this is truly the case, then our simulation is poorly built. Why would he say so? Because of how inconvenient it is: a universe that expands at an accelerating rate will make it difficult to understand the cosmos in a few hundred billion years, and if star formation drops at the current rate, the universe is doomed to darkness. Additionally, he points out that the peculiarities of physics "seem a bit patched together."

Now, I have a few things to say about all of this.

First of all, simulating stuff is hard. I mean, it's so much harder than you think. Summit, the most powerful supercomputer in the world has 9216 CPUs with 22-cores each, 27 648 GPUs and 262 144 000 gigabytes of storage space. To top it off, this beast requires a power supply of 13 million watts, which is roughly equivalent to 10 000 US households (or 32 000 German households, just saying). For comparison, an average home computer has only 1 CPU with 4 cores, 1 GPU, around 512 GB of space, and a power supply with 600 watts of juice.

Now this machine is certainly impressive, so what can it do? Well, it can simulate how a handful of atoms behave in nuclear fusion. That's pretty much it, to simulate something like a few grams of fuel inside a fusion reactor, we need a facility the size of two tennis courts and a weight of 300 metric tons. In terms of weight, you need a machine that is one hundred million times heavier than the thing you're simulating, and that's just for the low complexity stuff. The required computational power increases exponentially with increasing complexity and size of the simulation, and so does the size of the computer, not to mention the required computational time.

Although it's hard to know how powerful computers will be in the future, I think it's still safe to say that simulating some \( 10^{80} - 10^{90} \) particles of a whole universe is impossible within a universe of the same size (yes, even with quantum computers). Therefore, if such a simulation is made, it cannot represent the universe where the simulation hardware is. That's just how it goes, corners will be cut, must stay within the budget, cannot violate the laws of physics; so many restrictions!

Because of this, I support the idea that if our universe really is a simulation, then it has to be some kind of down scaled version of some other universe. As in, if we tried that with our current technology, we would end up with some kind of 8 bit universe (not the artist). So as you can see, the simulation hypothesis is not without problems. But wait, there's more!

If it really is likely that any given universe is a product of some type of simulation, then it is equally likely that the universe which is simulating us is also a simulation. This comes naturally from the "numbers game" motivation for the simulation hypothesis: if a large number of universes are simulated, then it is likely that any given universe is a simulation, regardless of whether that universe is also running a simulation or not. Therefore the universe simulating us is also equally likely to be simulated, and so is the next one, and the next one, ad infinitum. You just get an infinite chain of universes of ever increasing size, so it's turtles all the way down again. Or, Inception, if you prefer Hollywood over mythology.

Finally, let's take a closer look at the standard motivation. It was stated that "if there are a large number of simulated realities, it is very likely that our reality is also simulated," right? Frankly, this is a really stupid thing to say. When someone says this, they assume that the different universes are uniformly distributed within some probability space which contains all possible realities. The problems with this are that 1. it is impossible (with current knowledge) to know how these universes would be distributed and 2. whether the same probability space contains simulated AND non-simulated universes.

For illustrative purposes, I will now perform mind reading magic. Think of an animal; any animal is fine. Think of how it looks like, how it moves, what it does, and so on. Ready? You are thinking about: nematodes. No? Oh well, it was worth a shot. You see what happened there? I asked you to run a simulation - inside your brain - and then attempted to employ the "numbers game" idea to figure out what you were simulating. And I am pretty sure that you were not thinking about nematodes, more likely some domestic animal, like a dog, a cat or a horse, whichever animal you are most familiar with really.

By the logic employed by simulation hypothesis proponents, you should have been thinking about nematodes, since they are the most abundant animals in the world (they make up about 80 % of all individual animals on earth). We all know that this is not really how it goes, and this illustrates the first point. We don't know what the probability distribution is for being in a certain universe, and what variables may affect that. For all we know, there are infinitely many non-simulated universes, and only a finite number of simulations, in which case the "numbers game" logic is reversed.

As for the second point, consider for a moment that you are living in a non-simulated universe, but you yourself, don't know it. Now, someone somewhere makes a gigantic number of simulated realities. What are your chances of living in a simulation now? It is exactly zero, because your universe does not belong to the same probability space as the simulated ones! It simply does not matter how many simulations there are, if your reality does not belong to the same probability space, then it will not affect your chances of being in a simulation. But do we belong to the same probability space as simulated universes? Or do we group together with non-simulated ones? Or are they all in the same space? No one knows!

I personally don't see any good reason to assume that simulated universes are in the same probability space as non-simulated ones. And I certainly don't believe that it's turtles all the way down, that's just not science. In fact, the simulation hypothesis has more in common with religion, since religion can be defined as "a comprehensive worldview or 'metaphysical moral vision' that is accepted as binding because it is held to be in itself basically true and just even if all dimensions of it cannot be either fully confirmed or refuted."

Believing in the simulation hypothesis in itself is fine, but it's just not science. And as far as I can see, there is no good reason to believe that we are, in fact, in a simulation.