Jesse Mazer wrote,
> Isn't there a fundamental problem deciding what it means for a given > simulated object to implement some other computation?
Yes, but does this problem need to be solved? I have no problem with the idea that some "physical object" (in one computation) can be "interpreted" in all sorts of ways - depending on how you map it. Does it matter if there exists a (weird) mapping between a rock and a universe with conscious inhabitants? The universe doesn't depend on the rock for its existence so who cares!
I think it would matter if you want to find the measure of various types of observers/observer-moments--you need to know which ones are instantiated more often in the set of all possible computations (to address this you might also need a measure on all possible computations). Without some type of measure, there is no way to solve the "white rabbit problem".
Expand your wine savvy — and get some great new recipes — at MSN Wine. http://wine.msn.com