[ExI] Let's play What If.

[Damien Broderick]

On 10/27/2010 5:15 AM, Stathis Papaioannou wrote:

> torturing one of the copies would cause a differentiation, and
> there is a 1/million probability that you would be the copy that
> differentiates. On the other hand if there were only 2 copies there
> would be a 1/2 probability that you would be the one that experiences
> torture and differentiates. So it is better to go for the million
> copies.

The error here is that people are (mis)using the word "copy" when they 
mean "instance" or "instantiation."

The original of me (i.e. me here and now) is an instance of me; any 
effectively perfect copies of me will also be (at the moment of their 
creation) instances of me-as-I-was-when-scanned.

Bearing this crucial distinction in mind, and changing the rules so the 
torture can also be a risk for all the instances (since that seems 
important to you), the probabilities are actually 1/1,000,001 and 1/3 
(since you stipulated a million copies versus two copies).

Damien Broderick