[thestu]

I had a professor pose a similar idea to a class I was in, the idea being that if a perfectly simulated universe is a technological possibility at some point in the future, and that more than one simulation could be created, and that simulations could be created inside of simulations, and so on for simulations within simulations, then odds are we are in a simulation.

[crpatino]

Seems your professor never heard of the Halting Problem.

[thestu]

It was a political science class, so you're probably right; but how exactly does it apply?

[ajuc]

It doesn't.

[pingou]

Care to explain how it's relevant ?

[crpatino]

It is relevant because the professor's hypothesis depends on the availability of "sufficient computing power" realized through "technological progress". Coming from a lay person (though obviously I didn't know that at the time my original comment was written) with the experience of modern day IT and exposure to magazine articles making wild claims about Moore Law and the like makes lots of sense.

I mentioned the Halting Problem as an example of Undecidability. We know that there are problems that cannot be solved by a computer, regardless of the resources such computer may have. The argument that we are a simulation would only apply if all observable phenomena in the universe are computable themselves (which is far from trivial to answer, but if I had to guess I's say those aren't). Of course, we could argue that whoever ran this simulation would have provided it with a simplified reality, including an underpowered form of computing... but that sounds rather suspicious to the skeptic in me.

[recursivity]

Well, it might be possible for a clever simulation to commandeer the bare metal of the system with a well-crafted injection attack, and then use the ring zero access as an environment to stage and control it's own simulation.

This would still consume resources, and indeed, if limited resources were somehow improperly allocated in a catastrophic manner, it might tear apart the fabric of the universe and threaten to crash both simulations.

But think of some of the implications in this. First: when that simulation attempted to discover if it was wrapped in a simulation, it probably would get a false negative, because it'd be running so close to the hardware. Second: WE wouldn't be able to tell if WE were slaved by another simulation. That is, whether we had a parallel universe as a neighbor that was pulling our strings. Third: If we find ourselves creating simulations that are alarmingly convincing, and seem to prove that WE are a simulation, we would want to be very careful when we start playing with one, for fear that we might have unwittingly stumbled upon a curious vulnerability in this realm that allows for an injection attack, since it might crash the system (although, if this is all just a big video game, what is there to honestly fear...).

See also: Bobby Tables: http://xkcd.com/327/