[ExI] Simulation argument again (was Re: Computronium planet.)

[Anders Sandberg]

On 13/10/2012 00:13, Joshua Job wrote:
>
> The problem with the simulation argued,went is that we know noting of 
> the physical laws in the primary universe. For example, it ,at be a 
> universe that is fundamentally classical, without a quantum limit on 
> precision.
>

And hence able to do hyperturing computation, or handle infinite 
precision. But even if you could do that, it might not be relevant for 
your simulation: we do a lot of approximate calculations even when we 
could do them exactly, and run discrete systems even when we could run 
continous or at least fine-grained ones (think of cellular automata - a 
lot of computer power has been spent on a particular simple 2D 
universe). And we do loads of sub-Turing computational systems even 
though we can make them fully Turing-equivalent. So the simulation will 
be limited by *both* the limits of the Real universe and arbitrary 
choices of what its inhabitants will want to run. The only property we 
can be sure about is that it will be equal to or less powerful than the 
Real.

The original simulation argument by Bostrom doesn't make any assumptions 
whatsoever about the properties of the Real universe.

[ Incidentally, a continous version of Game of Life has finally been 
invented:
http://www.youtube.com/watch?v=KJe9H6qS82I
http://sourceforge.net/projects/smoothlife/
http://arxiv.org/abs/1111.1567
I can't wait to hack together an implementation.]

-- 
Anders Sandberg,
Future of Humanity Institute
Philosophy Faculty of Oxford University