>There is a selection effect by the very use of the suicide machine. In the
>usual WM experiment this doesn't occur, so let's modify it slightly.
>First I measure the z-component of a spin ½ particle is measured 1000 times
>in succesion. Provided I don't find 1000 times spin up I will perform the
>usual WM experiment, otherwise I will only make copies that end up in
>Suppose you use the suicide machine to select W ten thousand times in a row,
>what would the probability be that I had found 1000 times spin up?
Near zero. (A priori the W selection should not interfere with the spin
measurements unless you change the "goal" with the suicide machine).
Perhaps you are linking the experience by testing me on the WM after
linking your decision on the spin measurements. In that case I would
say the probability you found 1000 times spin up would be great, but
only by pure Bayesian reasoning (not linked to the duplication