Yesterday, I finished watching Plastic Memories.
I’m still kinda depressed about it, but it got me thinking about the feasibility of simulating a human brain or maybe even the entire universe. That’s how I arrived at this silly but somehow still logical hypothesis…
Imagine that we have a supercomputer used to simulate atoms. Now, let’s say this computer occupies a 1Km³ volume but is somehow able to simulate 1.1Km³ worth of atoms (of course, not all atoms occupy the same volume and some might be way harder to simulate than others but, just for the sake of this hypothesis, let’s say our computer can simulate most metals at a solid state).
Our supercomputer could then simulate a supercomputer slightly more powerful than itself, which, in turn, could simulate a computer slightly more powerful than itself, so on and so forth…
Which, of course, means INFINITE FREE COMPUTING POWER.
I know, I know… it sounds completely implausible and silly, but… apart from the assumption that our computer can somehow simulate more atoms than it is composed of, everything seems to make sense to me.
Of course, even at a much higher level of abstraction, perfectly emulating a CPU usually takes a CPU at least 10x as powerful as it. The thing is: the simulation doesn’t have to be perfect. We might be able to just simulate molecules or maybe even larger groups of molecules instead of individual atoms and use ideal molecule models instead of trying to simulate everything to the quantum level.
Also, we could neglect various aspect about the computer itself. For example, there’s no point in simulating the PSUs in each server, just assume the computers are somehow always powered on.
Well, anyways… I just wanted you guys to reality-check my silly idea…
Maybe our scientific research commissioner @Macaronee could point out its flaws?
P.S: Of course, eventually, the imperfections in the simulations would render one of the virtual computers unusable, but I think it would take at least a couple of levels of inception to make that margin of error be large enough to cause problems.
Besides, even if we could only simulate one virtual supercomputer, it would already be more than worth it.
Reminds me of the theory behind technological singularity