Firstly, simulating intrinsically chaotic weather does not have to be accurate. It's just has to be close enough to be indistinguishable from the real thing.
Secondly, weird rounding errors, randomness, aliasing effects, truncations etc. will be physical reality for the beings living in simulated world.
Thirdly, physical laws we encounter might not be the real physical laws. Imagine super quantum Newtonian universe (infinite light speed, infinite speed of causality) where relativity and quantum effects are caused by the limits of the computer architecture and are not the laws of the underlying universe.
I appreciate the comment, but we seem to have different understanding about the context of the article.
>It's just has to be close enough to be indistinguishable from the real thing
Then that 'close enough'simulation is the real thing for us. The summary of the content is, the computational complexity of that simulation is so large that we need a 'close-enough real thing' to simulate 'close enough simulation' of itself. In that case is it even a simulation?
The second & third point mentioned seems to have assumed that such computational complexity is achieved; it doesn't change my argument of it.
Firstly, simulating intrinsically chaotic weather does not have to be accurate. It's just has to be close enough to be indistinguishable from the real thing.
Secondly, weird rounding errors, randomness, aliasing effects, truncations etc. will be physical reality for the beings living in simulated world.
Thirdly, physical laws we encounter might not be the real physical laws. Imagine super quantum Newtonian universe (infinite light speed, infinite speed of causality) where relativity and quantum effects are caused by the limits of the computer architecture and are not the laws of the underlying universe.