At 15:49 -0700 23/07/2002, Hal Finney wrote:
>I took the liberty of copying a few paragraphs from James Joyce's
>book describing the causalist argument in Newcomb's Paradox. This is
>the best statement of the argument for taking both boxes that I have
>seen. I also included a short response of my own, which describes an
>alternate way of viewing the paradox based on multiverse models.
>It is at http://www.finney.org/~hal/Newcomb.html.
Why so much fuss for just having 1000$ more than 1000000$ ?
I would take box A, letting the 1000 dollars in box B as a pourboire for
the predictor :-)
A more symmetric version of the paradox prevents this joke.
The predictor gives you the choice between taking 2 boxes or just
one. In that case you can choose it at will. The predictor said
again that he can predict what you will do; and in case you will take the
two boxes, he will put nothing in it! If you take just one box, he
puts 1000000$ in each boxes.
This version makes possible to reason in term of Einstein's reality
elements (which he did introduce in the famous EPR paper).
Einstein defines an element of reality by the rule: If I can predict
with certainty the outcome of an experiment I could do (but will not do)
then the outcome I would have find corresponds to an element of reality.
The "rational" causalist Rachel reasons like that: the predictor is
very good (hypothesis) so by taking one box I will gain 1m$.
Now this does not depend on which box I choose. So I can predict with
certainty that if I take only box A, I will find 1m$ inside, and
similarly I can predict that if I take only box B, I will find 1m$ inside
too. Like in EPR the causalist Rachel supposes some form of locality
(no action at a distance, no action in the past, ...). This gives a
proof that the element of reality "owning 1m$" is true for each boxes.
So she decides to take the two boxes. And because the predictor is
indeed good, she wins O$!
The "irrational (?)" evidencialist Irene reasons like that: the predictor
is very good (hypothesis), so by taking just one box I will gain 1m$
---whichever box I choose! So let me choose just one box.
And she goes away with her box, and she wins 1m$. Rachel comes by and ask
her if she (Irene) realizes that she has just abandoned 1m$. 'No, said Irene.
'the predictor would have predict I would have take the two boxes, and he
would have put nothing in it. And I would have win 0$. Why not take seriously
the hypothesis that the predictor is good'. -'But still you know there
is 1m$ left in the other box', said Rachel. 'Yes, Irene said, 'but only
because the predictor knew I will take only one box. Now nothing prevent you
of taking the other box'. So thanks to "irrational" Irene, both of them
wins 1m$. And then they married and get a very happy life! The kind of
life you can get when you manage to handle both cause and evidence, a subtle
mixture of reason and madness perhaps!
My opinion: Giving the hypothesis that the predictor is good, I think
Irene makes the right choice. In both this version and the traditional one.
In real life, though, I would be doubtful that such a predictor can exist.
So I am not sure there can be a pragmatic content in those stories.
I think also Hal makes an interesting point showing perhaps the first (to
my knowledge) rational argument for a role of consciousness/free-will in
"collapsing" the wave packet(*), or for consciousness deciding the output of a
quantum experience, or, in this version, consciousness making up
elements of reality (all this deserves perhaps to be more deeply
The weakness of the argument comes from the existence of the predictor hyp.
(*) Still in Everett sense, not in the Copenhague sense.