Re: Craziness of a quantum suicidal

From: <hal.domain.name.hidden>
Date: Sat, 19 Jun 1999 17:00:47 -0700

Wei Dai, <weidai.domain.name.hidden>, writes:

> I think the main point of disagreement between the two camps now is
> relative SSSA versus absolute SSSA (Marchal's terms). Can we all agree
> that given the absolute SSSA, there is no justification for QS? (Higgo's
> "I don't care about anything" doesn't count.) If so, someone from the QS
> camp should respond to my argument that the absolute SSSA is necessary for
> practical reasoning.

(I see that my earlier attempt to respond to Wei's point about the
SSSA being the basis for practical reasoning was badly confused, but I
think I will have to learn more about the philosophical underpinnings
of probability theory before I can give a better response.)

I may have the wrong definition here; if so please could you briefly
summarize the absolute SSSA?

I think the idea is that each observer-moment has a measure which
describes the probability that it would be selected randomly from
among all observer-moments. Each moment of consciousness that we
experience is weighted by that measure, and it is much more likely that we
experience high measure moments than that we experience low measure ones.
In particular, our current experiences of reading and writing on this
mailing list should be relatively high measure compared to a potentially
vast number of extremely low measure and hence improbable ones.

The question then is whether we should commit quantum suicide (QS) in
this model. There are two problems that we have to dispose of first.
The first is that this universe is deterministic, hence questions of
policy may seem meaningless, since there is no free will. We will in fact
commit QS or not depending on the laws of physics and not any policies
we adopt. This objection I think we can dispose of in the standard ways
in which free will is coupled with determinism, mainly claiming that the
two phenomena are identical but occur at different levels of description.

The second problem is that there seems to be no need for a definition
of identity in this model; each observer-instant exists on its own,
and although the laws of physics are such that its measure will be
related to the measures of succeeding and following observer-instants,
there is no other necessary coupling between them. In particular we
don't assume that there is one person whose identity flows from one
moment to the next. Rather, each observer instant exists on its own.
So it is not clear what it means to ask whether "I" should commit QS,
if the concept of identity isn't used.

Nevertheless I think we can say that whatever policy is appropriate with
regard to QS can be expressed independently of any concept of identity.
The policy would be something like, "commit QS if circumstances at the
observer-instant in question satisfy condition X." The question is
then what rule a general observer should adopt in deciding whether to
commit QS.

We still need one more element, though: what the goal is of the observer?
He needs to decide whether to commit QS based on some criteria.
What should they be?

It seems to me that we can identify two goals which are both plausible,
one which would justify QS and one which would not. The first would
be to maximize the average happiness of all observer-instants in the
universe. The second would be to maximize the total amount of happiness
in the universe. (I believe this was discussed earlier on the list,
so I apologize for going over old ground, but I have forgotten the
resolution if any.)

In the first goal, if we look at a transitional observer-instant,
suppose we have that the happiness level going into that instant is 100.
Then the universe splits into good and bad branches, each with probability
.5 compared to the original branch. In the good branch the happiness
level is 120, and in the bad branch it is 80. In this little piece of
the big picture, the average happiness level is 100 (100*1 + 80*.5 +
120*.5)/(1+.5+.5).

If QS is used to prune the bad branch, the average happiness level climbs
to (100*1 + 120*.5)/(1+.5) or 106.7. So in this case QS would appear to
be justified. All observers should adopt the policy of committing suicide
whenever something bad happens, which will raise the average happiness
lvel of all observer-instants in the universe.

If the second goal is adopted, then QS is not justified. Even bad things
have a positive happiness value. (If things get so bad that there is a
negative happiness value, suicide might be justified, but that would be
true in a single universe model as well so this does not count as QS.)
In the situation above, the total happiness is 100*1+80*.5+120*.5 or 200,
while if we commit QS the total happiness drops to 160.

So it seems to me that even with the absolute SSSA (if I have understood
it properly) there is a matter of goals which must be resolved before
we can say that QS is unjustified.

Hal
Received on Sat Jun 19 1999 - 17:04:40 PDT