Indeed, what I find much more fascinating is that these ideas are not new at all, but each generation frames it in a new technological light.
It's like how the brain used to be an ocean, then it was a windmill, then it was a steam engine, then it was a computer, and now it's a GPU cloud. I wonder what will come first: humans finally understanding what a brain is, or making one.
SF authors have explored these fascinating concepts without falling into such holes in logic: such as presuming that a technology capable of simulating an incomprehensible volume of intelligence was simultaneously too lazy to leverage that intelligence to not cut corners in the simulation.
That said, I do subscribe to the school that says we must conclude one of the following:
a) civilization in the abstract generally does not progress to the point of being able to simulate itself, or
Given our experience with actual simulators, the fact is that simulating a world populated with intelligent beings capable of simulating another world, going into a recursive loop, requires infinite processing capacity.
My guess is that simulation indistinguishable from reality is never going to happen because it isn't possible, having nothing to do with how evolved we are.
But from the inside of the simulation reality is whatever the rules of the simulation say it is. Thus the observer in the simulation has no way to know whether the reality he observes corresponds to an objective reality that exists "out there" or he's living in an arbitrary reality. For example, we can create game worlds in two dimensions, and those worlds are in a sense internally consistent, but not representative of our reality.
> Given our experience with actual simulators, the fact is that simulating a world populated with intelligent beings capable of simulating another world, going into a recursive loop, requires infinite processing capacity.
This reminds me of old-timers talking about "You'll never use 640 kB of RAM!" It is rather odd when thinking about "infinity" though, but there are a couple things that I would note: 1) It's hard not to accept the concept of infinity. It's used quite a bit. 2) In order for one infinity to exist, all infinities must exist. 3) http://phenomena.nationalgeographic.com/2015/04/24/when-hubb...
Those are galaxies...in a tiny patch of "empty space". Space is big. Really big...
The simulation doesn't have to run in full detail, there could be approximations made far away from sentients. Also, the speed of the simulation doesn't need to be 1:1.
> Also, the speed of the simulation doesn't need to be 1:1.
This is what I think of when considering spacetime and relativistic effects. If you think of "matter" as being equivalent to processing power required, then the more processing required, the longer it takes to update the entire frame. So other things with less matter are updated more quickly relative to more massive segments.
So what you're saying is that to engineer a system where clients pay me to do work for them, all I need to do is engineer a system where clients pay me to do work for them?
It's like how the brain used to be an ocean, then it was a windmill, then it was a steam engine, then it was a computer, and now it's a GPU cloud. I wonder what will come first: humans finally understanding what a brain is, or making one.