The stalwart Kimota94 posted this little gem about that philosophy paper making the rounds on the ole inter-weeb. You know - this one - that claims that this may all just be a simulation. Kimota94 does a good job refuting that idea with an excellent argument. He also asks "if this is a simulation, what would you do different?"
What would I do different? One argument would say "nothing - it's not possible to do things differently than the simulation would predict." Another would say "nothing - the simulation theory is nice but doesn't really impact me." A third would be "nothing - the amount of computing power necessary would require more operations per second than atoms in the universe so it would be unlikely to reach that level of computational power."
I'll start with the second one: it doesn't impact me. Even if I were to exist within a simulation, what I perceive as a "mind" is what I perceive. As a corollary to self-awareness, that would me the simulation created life (me) and I'm operating within it. Ipso-fatso, I am the simulation, but it is no different than what I was before.
The first answer justifies my view that "what would you do differently?" is the wrong question. I think I would rather ask "what would you simulate if you were in charge?" That would be interesting. The simulation described would have the power to simulate any possibility so changing your behaviour would be accommodated. Setting up different startup conditions or end conditions and running the simulation would be more interesting.
Another interesting aspect of the paper that I didn't see addressed was the time scale of the simulation. Would the simulation be in real-time or at some accelerated rate? I would hope that it would be accelerated as real-time would be a bit of a drag. I suppose if it is accelerated than we, as the simulacrums are thinking faster than the items being simulated. If indeed we are simulated, then is every single possible activity being simulated, or are short-cuts being taken? Computer modeling (as I know it) is discrete and therefore leaves out some information in the interest of producing a useful model. If the proposed machine is not modeling, then it really is new life in a novel media - AI if you will.
Finally I would say ('cause I really ought to be getting some sleep now) that I do not think such a model is possible. Let's take one of the better known systems used to model the world around us - mathematics. It has been demonstrated that any mathematical system is incomplete. As a corollary, the simulation system described would be incomplete, no matter its power. It would not be able to simulate life exactly, by the same argument. Therefore the simulation would be something different than the previous existence it is attempting to simulate, which implies that such a simulation would not be stable. It would crash soon after starting. Maybe it is a new religion - we are all just waiting the Great Blue Screen of Existence. The Reboot is Near - REPENT! Yes, sleep is needed. By me - maybe you, but definitely me.