[as300]

> a computer to have sufficient accuracy to model the world would by definition need to have as much memory as the world, and model itself in it.

But what if we had certain "compression" abilities that allowed us to simplify the world? Similar to how storing an audio codec lets us massively minimize how much storage we need for music?

[keerthiko]

Magical compression algorithms that have amazing compression ratios and amazingly computational cost to reasonably decompress to process the data could theoretically raise the upper bound of a computational world. But compression cannot solve for an unbounded recursion problem in the simulator having to model itself it's model of the world, unless we arrived at an analytical closed-form solution for any prediction rather than relying on any modeling.

[neuronexmachina]

Or use lossy compression and get quantization noise. The model would probably get pretty weird if you looked at a fine enough detail. ;)