OK, first imagine its a computer simulation. This is fine with your definition. Then imagine that the simulation is distributed across servers around the world. Still OK?

Now imagine that we distribute it to servers across the universe... Still OK?

So the data that represents the sim is effectively scattered over the entire universe, and if you were to look at how it exists physically it would seem to be bits of random data (in the absence of the "program" which knows how to interpret that data)

Now imagine that we slow down the simulation program. If you were in the Sim you would not notice any difference. We can speed it up too, but time in the universe is irrelevant to the Sim, as long as the next calculation is done to create the data for the next step for the Sim.

So really we could even simulate them out of sequence (if this was programmatically possible) and the Sim would not notice.

Now what if nobody looked at the computer to see if it was working. Would that be OK?

What if it was never looked at?

So the only thing that "knows" the simulation is the computer. Is that OK?

The computer is just more data. More atoms of data storage which are considered to represent something. So why have the computer at all? What if there was no computer and the data representing the simulation steps just happens to exist. After all it was only the computer that required it to be in specific places in a specific order, and we just scattered them over the whole universe anyway. For example, say part of the data is where the computer expected it to be, well we could just have that data somewhere else - after all there is now no computer to "care" where it is. Or we could just consider some other random atoms to represent the computers new record of where that data should be.

So when above did the "simulated" reality stop existing, because now its there and there is no computer, nobody looking at it and it is made up of matter spread over the whole universe. Why not all of the matter?

Approach 2, is my conceptual one. Which is to point out that if it can be simulated then each step in the simulation (or if you like the entire history of the simulation from beginning to end) is just data. So as data we could represent it as a very very very large number.

That number may not be represented physically, but does that stop it existing? It very likely exists within the digits making up pi (you are almost certain to find the first or last 6 digits from your phone number in the just the first 100million digits) imagine what there is within the infinite number of places it has. Pi is a thing that exists and is represented by every circle. Pi is just one of many conceptual places that a reality could hide.

I don't realy see a contradiction here, with what i said.
But i have two questions: when does a parallel reality become more than just a parallel reality? because you're basically saying that there could be an infinite amount of parallel realities hiding, everywhere...wich is a thought i realy like btw. But yeah, what's the difference between interconnected parallel realities and a simulation?
And secondly, wouldn't you still need at least a decoding mechanism somewhere? Something isn't realy a code, unless there's a decoding mechanism of some sort, even if it would be merely a theoretical decoding mechanism. If you even on a theoretical plain, could no longer decode a message, the message is lost, isn't it?

What is Musks reasoning behind this assumption? Is it that there would have to be infintely many of these hyper-advanced civilisations, and therefore infinitely many simulations?