Bruno Marchal, <marchal.domain.name.hidden>, writes:
> Here is the message of Nick Bostrom where he introduces briefly
> the SSA (in this discussion list 13/04/99):
>
> >The Self-Sampling Assumption (SSA), the idea that you should reason
> >as if you were a random sample form the set of all observers,
> >underlies many of the discussions we have had on this list. About
> >half a year ago I discovered some paradoxical consequences of this
> >assumption. It seems to imply that weird backwards causation and
> >psychokinesis(!) is feasible in our world. In this small paper I
> >describe these possible counterexamples and discuss whether they
> >really are as paradoxical as they appear at first blush:
> >
> >http://www.anthropic-principle.com/preprints/cau/causation.doc
Perhaps we need to distinguish a "Strong Self-Sampling Assumption",
which is like the SSA but instead of discussing "observers", it refers to
"observer-instants". An observer is somewhat ambiguous, not so much in
the sense that it is hard to say what counts as an observer, but rather
because it seems to implicitly bring in the concept of identity.
I am the same observer today that I was yesterday, and will be tomorrow.
To say that I am a random sample among all observers is to refer to my
lifetime of experience as a single instance of an observer.
The Strong SSA, on the other hand, refers not to observers, but to moments
or instants of observation. At each moment we have to view ourselves
as a fresh, newly selected instance among all possible observers.
Of course, our moment of experience does include memories of a lifetime
of past experiences, but that is not relevant to the definition of an
observer-instant in this concept. There is no necessary equality of
my probability of being an observer-instant now, and the probability of
the observer-instant which corresponds to my memories of one second ago.
This is especially true when we begin running our exotic thought
experiments of quantum suicide and mental cloning. If I sit on a bomb
and press the firing button, and it doesn't go off, it may still be
the case that the probability of my current observer-instant is much
less than the probability of the one just before I pressed the button.
If I step into a duplicating machine which makes 100 identical and
indistinguishable copies of me, the probability of my observer-instant
after the event (but before we have begun to diverge) may be much greater
than the probability just before the button was pushed.
I think that Nick Bostrom, who (as far as I know) conceived of and named
the SSA, was not thinking in terms of the Strong SSA. When he discusses
"my" birth rank, he is implicitly assuming that I am the same observer
today that I was when I was born. I think it is best, therefore, to
distinguish the two assumptions by giving them names in this form.
Hal
Received on Tue May 18 1999 - 09:38:42 PDT