RE: PhD-thesis on Observational Selection Effects

From: Bostrom,N (pg) <"Bostrom,N>
Date: Thu, 27 Jul 2000 23:00:29 +0100

Jacques Mallah writes:

>If I remember correctly, I was not convinced that MWI
prevents these

                I though you were. In you say: [snip] You also
said in that post you weren't sure how to combine the QM probabilities with
the SSA, but in your last post to RS you seemed reconciled to doing so.

Consider first a partially analogous case, where we assume MWI is false.
Suppose there are a trillion planets (analogous to different branches on the
wavefunction), and an Adam and an Eve doing the experiment on each of these
planets. On the vast majority, no deer turns up, but on one in ten thousand
planet, the trick works. Now, if Adam knows that this is the setup, then he
should not think that the trick will work. The reason is that the no
outsider requirement is not satisfied: to apply SSA with a reference class
consisting only of the humans living on his planet would be wrong. Once we
take the reference class to contain all these actual observers living on
some planet or other, then it is easy to see that applying SSA will not give
Adam any reason to think that a wounded deer will appear.

If this is a good analogy to the MWI-case, then the paradox would not arise
under that assumption. But that is not clear to me at all. For the branches
of the wavefunction aren't simply so many different actual worlds, but they
come with a measure. I take it that we need to assume that this measure is
related to SSA by assigning sample densities to observer-moments that is
proportional to the measure of the branches on which they are living. This
means that observers living on small-measure branches will only weakly
contribute to the violation of the no outsider requirement; that is, if one
of the alternative hypotheses under consideration is only true on branches
which have a small measure, then the no outsider requirement can for all
practical purposes be regarded as satisfied.

Therefore, if Adam is uncertain whether his world is one where the
deer-branch has a high or a low amplitude (but thinks that with 99,99%
probability it has a low amplitude), then when he considers his position, he
gets reason to think that the deer-branch has high amplitude, since that
would concentrate a greater sampling weight on observers that are observing
what Adam observes. (On the other hand, if Adam were certain about the shape
of the wave function, then the above argument would not show that he should
think that he will next observe a wounded deer. But why assume that Adam
knows the shape of the wave function? If he doesn't, then it seems to me
that my paradox obtains also in the context of the MWI.)

>Even if it did, I don't think it would be very relevant
>we are concerned with what is a sound methodology,
something which, one > may
>think, should be independent of specific assumptions about
the world. In
>particular, If one is advocating a general principle (such
as SSSA-R) then
>a single genuine counterexample, even if it is
hypothetical, would suffice
>to refute it or force us to restrict its scope.

                    I'm not sure what you mean in this context.

I mean that few people would think that SSSA-R stands or falls with MWI.
Just as it is no good reply to the Adam&Eve paradox that Adam and Eve
probably never existed, so it also seems to be unsatisfactory to dismiss the
paradox by saying that it doesn't obtain if MWI happens to be true (even if
that were true, which for the reasons above doesn't seem the case).

                 The DA is not counterintuitive as far as I am concerned.
It's common

Not very common, in my experience.

                    When I said SSSA I meant SSSAR.


                    In the 'coin toss' case, the answer is obvious.

Yes, but different answers appear to be obvious to different people.

>The main motivation is that it avoids certain consequences
that most people
>regard as highly counterintuitive and unacceptable.

                    Well, I've already stated that I regard those
consequences as obvious
                and intuitive.

So you accept that Adam should think that a wounded deer will turn up?
(Let's assume here for the sake of the argument that MWI is false.) You
would probably be the only person who thinks that that is obvious and

                    I would follow any advice I would give to Adam if I were
in his

It's quite interesting to know that there is someone who would do that. I
though I would for a while, but after thinking more about this and making
the situation vivid to my imagination, I came to the conclusion that this
seemed crazy and that if there were any consistent way of avoiding it, then
it should be avoided. Of course, everything I say in the thesis is
compatible with accepting R=Ru (i.e. setting the reference class equal to
the class of all observers) provided one is also willing to accept the
Adam&Eve paradox.

Nick Bostrom
Dept. Philosophy, Logic, and Scientific method
London School of Economics
Homepage: <>
Received on Thu Jul 27 2000 - 15:11:58 PDT

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:07 PST