Because the smallest simulation of something incompressible is itself. Assuming the universe is incompressible, it's impossible to simulate the universe in the universe. The smallest computer to simulate the universe would be have the size and energy requirement of the entire universe itself.
This example is great. My main point is that due to the laws of their universe, they will never be able to simulate any area of it in a smaller area, because the Game of Life isn't compressible.
What evidence is there that physics is 'compressible' rather than simply 'uncertain'?
Belief that consciousness is somehow more fundamental than 'brain physiology' is essentially a kind of solipsism and is completely unfounded in facts or experiments.
If we're in a sim, it's Physics rules that are being "run" and the existence of brains is purely emergent.
It's not a trivial assumption. I actually lean towards an incompressible universe, intuitively.
If the universe is incompressible, then you may only ignore parts of the data for a finite time. For example, arranging an illusion of the night sky might have taken no computation before astronomy, but as we record the positions of celestial bodies and understand the laws governing them, the computations have to be run retroactively since the big bang. Otherwise the simulation would be inconsistent.
So the question is what can you remove from the causal chain from the big bang to this moment? If you think some things are removable, you have to prove that something exists in the universe but does not touch any causal chain humans (or perceivers) may wish to investigate at any point in the future. If nothing is removable, then you cannot choose to not simulate any part of reality based on perception, because whatever you don't simulate may eventually need to be simulated.
There's no reason to think those computers don't exist.
We have no idea what constraints those machines would or wouldn't have, since we have no idea what physics they'd be operating under.
I'm only arguing against partial simulation within our own universe (eg. The Matrix). Any simulation that has a night sky and telescopes must simulate the full big bang and evolution of galaxies just for the people on Earth.
If it's the same, perhaps it's shown to us as a fraction of what's real.
However I think whether a simulation of only the perceivable universe is not possible, which means you can't reduce the required computation. See my reply to a sibling comment.
If the universe is incompressible then it's definitely impossible to simulate it within itself.
Perhaps the far universe we perceive and will perceive is not there anymore. It doesn't have to. What we perceive is its light (or electromagnetic/gravitational waves). Objects far away could have been garbage-collected millions of years ago. You don't need to compute them anymore.
If the simulation has a limited time to run, then you can create and delete objects while playing with the speed of light to carry whatever information to us.
Edit: millions of years ago is not needed. They could have garbage-collected the objects as soon as they complete the time for duration of the simulation. Whatever it is, I would save this information as a template for future simulations.
That's significantly larger than Earth, only to simulate Earth today, because tomorrow the entire universe would go dark.
Lightspeed gives structure to the computation of the universe but it doesn't reduce the load.
Saving templates doesn't reduce the load either, not in the sense I'm considering. It may practically mean that you don't have to run it again, but theoretically if you want to know how much computation got you to this point then you have to count the computation of the templates.
So if you had our whole universe at your disposal, and knowledge of all natural laws and processes, and knowledge of the starting conditions, and absolute power over matter and energy, you still cannot simulate our entire universe with anything less than just restarting it. If you wanted to simulated it specifically for some observers, you can free up resources using the lightspeed method but that's a computational debt you cannot pay if you wanted the simulation to continue.
> Lightspeed gives structure to the computation of the universe but it doesn't reduce the load.
The load would be saved because you don't need to keep computing interactions between object, because we don't get to see these object/interaction since the simulation has a limited time. You save this pre-render of the universe and you just need to calculate whatever is closer to us (and our simulated world).
You don't need to calculate all at once and you don't need to calculate since the big bang, since we will experience a very short period in the simulation. And I suppose most of it can be just faked.
Imagine a spherical universe with the Earth at the center. You can calculate the outer portion of the sphere, then garbage-collect, then the result is added to the calculation for an inner portion, then garbage-collect, an so on until you reach the center of the sphere.
Another thing is the time-constant. Why our time has to pass with the same speed as the time outside the simulation^? What we think it takes time to calculate it's now completely relative. If time is infinite in the real universe, then they have plenty of time to build our world so effectively that our entire existence is ran in a fraction of a second.
Other aspect is why our universe has to be of the same quality of the real universe. Perhaps we have a sloppy version that is easier to calculate (but it's still complex for us to understand).
Of course, I'm just drinking my coffee and playing I am a genius this evening: I'm not going to defend the above as in a real analysis.
Edit: ^changed calculation for simulation.