Saibal Mitra wrote:
>I think that your example below was helpful to clarify the disagreement.
>You say that randomly sampling from all the files is not 'how real life
>works'. However, if you did randomly sample from all the files the result
>would not be different from the selective time ordered sampling you
>suggest,
>as long as the effect of dying (reducing the absolute measure) can be
>ignored.
You would have to follow the branching pattern as defined by the program. I
suggested that one file is chosen at random from each folder and a new
folder generated from this file. It would be very resource-hungry, but it is
simple enough to imagine the computer generating the entire multiverse, i.e.
each file branching out to a new folder, each file in each new folder
branching out to another new folder, and so on. This would generate all the
OM's associated with an individual. However, you could just take any files
from any folder and cobble together an individual history, which is what
would happen if you sample files at random. To create an individual history,
you have to trace a single path through the (perhaps infinitely) branching
tree, which is a very different matter. This is true even if you ignore the
effect of dying, so each folder/time slice has the same number of files/same
measure.
>If I'm sampled by the computer, I'll have the recollection of having been a
>continuum of previous states, even though these states may not have been
>sampled for quite some while. I'll subjectively experience a linear time
>evolution. The order in which the computer chooses to generate me at
>various
>instances doesn't matter.
It does matter in that the tree structure described above, and the path of
an individual history through the tree structure, is dependent on doing the
computations in order. In a trivial sense you could say that the order does
not matter in that if the computation were predetermined in some way, then
the computer could generate the folders and files in any order.
>There are a few reasons why I believe in the ''random sampling''. First of
>all, random sampling seems to be necessary to avoid the Doomsday Paradox.
>See this article written by Ken Olum:
>
>http://arxiv.org/abs/gr-qc/0009081
>
>
>He explains here why you need the Self Indicating Assumption. The self
>indicating assumption amounts to adopting an absolute measure that is
>proportional to the number of observers.
I'll look that up eventually. The SSA, anthropic principle, Doomsday
Paradox, etc., I admit give me headaches. I think there is something
basically wrong or confused with these ideas and am hoping that one morning
I will wake up and realise what it is. Until that happy day arrives I will
have to confine myself to answering that if "random sampling" saves us from
the Doomsday Paradox, then maybe we are all doomed, because random sampling
across all possible OM's in the multiverse cannot possibly give a coherent
individual history.
>Another reason has to do with the notion of time. I don't believe that
>events that have happened or will happen are not real while events that are
>happening now are real. They have to be treated in the same way. The fact
>that I experience time evolution is a first person phenomena.
Fair enough. You can consider the tree as described above drawn timelessly
on the fabric of reality. You still have to take a subset of files/OM's from
the tree in such a way as to define a single path through each node in order
to give the first person impression (you could say: illusion) of a single
individual history in time.
>Finally, QTI (which more or less follows if you adopt the time ordered
>picture), implies that for the most part of your life you should find
>yourself in an a-typical state (e.g. very old while almost everyone else is
>very young).
That's perhaps right, although it is not without problems, when you consider
that being 30 or 40 years old is no more atypical when you expect to live to
infinity than being millions or trillions of years old. If it is right, at
least in the prediction of your age relative to everyone else, then maybe it
is an argument against MWI?
--Stathis Papaioannou
>Van: "Stathis Papaioannou" <stathispapaioannou.domain.name.hidden>
>Aan: <everything-list.domain.name.hidden>
>Verzonden: Monday, May 30, 2005 04:02 PM
>Onderwerp: objections to QTI
>
>
> > I thought the following analogy might clarify the point I was trying to
>make
> > in recent posts to the "Many Pasts? Not according to QM" thread,
>addressing
> > one objection to QTI.
> >
> > You are a player in the computer game called the Files of Life. In this
>game
> > the computer generates consecutively numbered folders which each
>contain
> > multiple text files, representing the multiple potential histories of
>the
> > player at that time point. Each folder F_i contains N_i files. The first
> > folder, F_0, contains N_0 files each describing possible events soon
>after
> > your birth. You choose one of the files in this folder at random, and
>from
> > this the computer generates the next folder, F_1, and places in it N
>files
> > representing N possible continuations of the story. If you die going
>from
> > F_0 to F_1, that file in F_1 corresponding to this event is blank, and
> > blank files are deleted; so for the first folder N_0=N, but for the
>next
> > one N_1<=N, allowing for deaths. The game then continues: you choose a
>file
> > at random from F_1, from this file the computer generates the next
>folder
> > F_2 containing N_2 files, then you choose a file at random from F_2,
>and
>so
> > on.
> >
> > It should be obvious that if the game is realistic, N_i should decrease
>with
> > increasing i, due to death from accidents (fairly constant) + death
>from
> > age related disease. The earlier folders will therefore on average
>contain
> > many more files than the later folders. Now, it is argued that QTI is
> > impossible because a randomly sampled observer moment from your life is
>very
> > unlikely to be from a version of you who is 1000 years old, which has
>very
> > low measure compared with a younger version. The equivalent argument
>for
> > the Files of Life would be that since the earlier files are much more
> > numerous than the later files, a randomly sampled file from your life
>(as
> > created by playing the game) is very unlikely to represent a 1000 year
>old
> > version of you, as compared with a younger version. This reasoning
>would
>be
> > sound if the "random sampling" were done by mixing up all the files, or
>all
> > the OM's, and pulling one out at random. But this is not how the game
>works
> > and it is not how real life works. From the first person viewpoint, it
> > doesn't matter how many files are in the folder because you only choose
>one
> > at each step, spend the same time at each step, and are no more likely
>to
> > find yourself at one step rather than another. As long as there is at
>least
> > *one* file in the next folder, it is guaranteed that you will continue
> > living. Similarly, as long as there is at least *one* OM in your future
> > which represents a continuation from your present OM, you will continue
> > living.
> >
> > --Stathis Papaioannou
> >
> > _________________________________________________________________
> > Meet 1000s of Aussie singles today at Lavalife!
> > http://lavalife9.ninemsn.com.au/
> >
>
_________________________________________________________________
Don’t just search. Find. Check out the new MSN Search!
http://search.msn.click-url.com/go/onm00200636ave/direct/01/
Received on Tue May 31 2005 - 00:09:36 PDT