Le 04-janv.-06, à 02:37, Stathis Papaioannou a écrit :
The problem of gradually fading away can be illustrated by another example. Suppose your body is destructively scanned and then reconstituted in two separate locations, a1 and a2. At a1, the reconstitution goes as intended, but at a2 something goes wrong and you are reconstituted in a brain dead state. I think we can say in this case that you can expect to find yourself alive at a1 with 100% certainty a moment after you undergo the scanning. Next, suppose that after the destructive scanning your body is reconstituted in 10 different locations, b1 to b10. As before, at b1 the reconstitution is perfect and at b10 something goes wrong and you are reconstituted in a brain dead state. At locations b2 to b9, however, due to varying degrees of malfunction in the machinery, you are reconstituted with varying degrees of dementia: at b2 you are just a little bit more vague than usual, at b9 you are still alive but have lost all your memories and sense of identity, and in between are several variations with partial dementia. The question now is, when you undergo the scanning process, should you have an equal expectation of ending up at each of the locations b1 to b10? If you exclude b10 because you are dead there, should you not also exclude b9, where you are no longer a sentient being, let alone a particular sentient being? And does it follow from these considerations that you are are somehow more likely to find yourself at b2 than b8, for example?
Interesting and hard question. I would say "intuitively" that all what matters are the infinite branches. If you fade away in such a way that some of your next observer-moments will lead to a dead end, drop it from the probability calculus.
Now to ask this to a lobian machine is quite another story ... Bruno http://iridia.ulb.ac.be/~marchal/