« March 13, 2006 | Main | March 16, 2006 »

Wednesday, March 15, 2006

Physics: Simulations All the Way?

Several people have sent feedback from The Cosmic Landscape review recently posted here to inquire whether I'd considered the possibility that, if our universe was indeed a simulation, there might not be a top-level (non-simulated) universe at all, but rather a closed loop (or infinite regress) of simulations. Yes, I have considered that possibility, but there's only so many impossible things you can squeeze into one book review! Here are a few comments about the approach which can be summarised as:
There is no top level; it's simulations all the way!
or, alternatively,
It from bit,
Bit from it.
Round and round,
Never quit.
I haven't thought about this in great detail, but it seems to me that the “simulations all the way” model may be phenomenologically isomorphic to both the eternal inflation/string landscape and Lee Smolin's cosmological natural selection (CNS) (original paper) models. If the creator of the simulation picks the parameters at random, it's like the string landscape, and if he twiddles them slightly from the values in his own universe it's like CNS. But in any case there is an unavoidable natural selection component because only universes which permit the existence of beings able to create simulations will beget simulated progeny. Hence, if we are in a simulation, it is descended from an unbroken line of habitable simulations.

As to the closed loop of simulations, it sounds crazy, but if the amount of state in a simulation is finite, then the possibility of a loop kind of falls out of the Poincaré recurrence theorem; whether there was a top level or not is unknowable once you've passed through sufficient generations of simulations to explore all possible initial states and transition rules. This is much like Tegmark's "Level I" parallel universes, where an infinite number of Hubble volumes containing all possible initial conditions exist.

Posted at 20:35 Permalink