>I must have misunderstood the concept of ``doppelbrothersisters´´. I meant
>that identical copies should be identified.
I have nothing against that identification, but I guess you agree we should
stop the identification when the neighborhoods differs so that the copies
begin to diverge.
I mean identical copies embedded in identical comp histories can perhaps be
identified. We need some prudence because a literal identification
could alter the "correct" relative probabilities.