Only a mystery for people who don't know anything about thermodynamics or information theory.

Here's the simple disproof: random phenomena. It turns out that random phenomena can be modeled mathematically as to their implications, such as the kinetic theory of gases, which is very exact. The point is: a computation cannot produce a random series. We have been able to produce only pseudo-random series. If we were in a simulation, we would observe deviations from the results predicted by the assumption of a random process---but we do not see any. (This is basically the same disproof used against the argument that there are Hidden Variables that determine "random" phenomena.)

For those who think some imaginary masterminds could produce a computed random series, they have some problems. How do you start the computation? Do you get the same result from identical starting conditions? You have to, if it is an algorithm. Which means that the series embeds the information contained in the algorithm. But a random series has no information, so there will be a difference in entropy between a random series and an algorithmic series. Where is the missing entropy?

Learn more. Fantasize less.