On Sun, Dec 09, 2001 at 08:51:04PM -0500, Nick Bostrom wrote: > Suppose we accept this view. The phrasing of the simulation-argument would > then change to one where the simulation-hypothesis states that the vast > majority of machines running your mind are virtual machines. I would then > argue that our preferences are such that if we think that most of our > implementations are virtual machines then we should care most about the > consequences of our actions for this majority of implementations (just as > in MWI we'd want things to go well for us on branches that get higher > measure). Then I think the practical implications of the argument would > remain the same.
I much prefer this way of phrasing the argument, as it makes things clearer. Thanks for the rephrasing. I made a case for this style of reasoning earlier at http://www.escribe.com/science/theory/m2462.html. One thing this rephrasing makes clear is that the argument depends on assumptions about people's preferences. Why do you believe that "our preferences are such that if we think that most of our implementations are virtual machines then we should care most about the consequences of our actions for this majority of implementations"? In MWI it makes sense to assume that we'd want things to go well for us on branches that get higher measure, because people who don't could most easily achieve their goals through quantum suicide and so we wouldn't see very many of them. However this argument doesn't apply here to the simulation argument. > Hmm. Would you say the analogous thing about temporally distinct but > qualitatively identical observations of the same person? Say, you wake up > first at 7 am for one minute and again at 8 am for another minute. Suppose > you know all along that you will awake on these two occasions and that both > times you will be in the same conscious state of thinking "It is 7 am now." > It would seem natural to say that the first time you awoke, you were right > and the second time you were wrong. Suppose a little later you look at the > clock and it says "8:01". Mightn't you then say to yourself "Gee! When just > a second ago I just thought it was 7 I was badly mistaken!" I would say that I was wrong both times, in the sense that both instances of the belief "It is 7 am now" are unjustified. The correct belief is "It may be 7 am or 8 am now", or "There are two brain instances implementing my thoughts, one at 7 am and another one at 8 am."