Le 15-déc.-05, à 03:04, Saibal Mitra a écrit :
To me it seems that the notion of ''successor'' has to break down at
cases
where the observer can die. The Tookies that are the most similar to
the
Tookie who got executed are the ones who got clemency. There is no
objective
reason why these Tookies should be excluded as ''successors''. They
miss the
part of their memories about things that happened after clemency was
denied.
Instead of those memories they have other memories. We forget things
all the
time. Sometimes we remember things that didn't really happen. So, we
allow
for information loss anyway. My point is then that we should forget
about
all of the information contained in the OM and just sample from the
entire
set of OMs.
The notion of a ''successor'' is not a fundamental notion at all. You
can
define it any way you like.
?
It will not lead to any conflict with any
experiments you can think of.
?
Counterexamples will appear if I succeed to explain more of the
conversation with the lobian machines.
But just with the Kripke semantics we have a base to doubt what you are
saying here. Indeed, it is the relation of accessibility between OMs
which determine completely the invariant laws pertaining in all OMs.
For example, if the multiverse is reflexive the Bp -> p is true in all
OMs (that is, Bp -> p is invariant for any walk in the multiverse). If
the mutliverse is "terminal" of "papaioannou-like) then Dt -> ~BDt is
a law. In Kripke structure the accessibility relation determined the
invariant laws.
later, the modal logic is given by the machine interview, and from
that, we will determine the structure of the multiverse, including the
"observable" one.
Bruno
http://iridia.ulb.ac.be/~marchal/
