[extropy-chat] Improvements to Newcomb's Paradox
lcorbin at rawbw.com
Fri Apr 6 02:17:49 UTC 2007
Newcomb's Paradox has only one admissable subject
behavior: take just the one box.
An evolutionary proof of this is as follows: suppose that
box A either does or does not contain dinner, and box
B always contains dessert. One may live on a steady
diet of box A, but one gradually dies of nutritional
deficiencies by selecting only box B.
(Those familiar with Newcomb's Paradox should skip
the present paragraph. The AI, or Deity, or whatever
entity has an unassailable track record of always
seeming to know whether you will always take just
box A or whether you will succumb to the temptation
to take both boxes. In some accounts---like wikipedia's
---it is "box B" which contains the necessary (or extremely
desireable) reward, but in other accounts that box is
Thus evolution would suggest that taking just one box
is an ESS. Or, as Jef Albright would say, taking only
the one box works.
In my own essay, I provide reasons why we should
regard "changing the past" as eminently reasonable
from the point of view of the subject. (See
Note the analogy to the free will discussion that we have
been having. If you imagine an audience, especially one
composed of physicists (and--better yet--physicists who
love you and want only the best for you), then as the
Alien made his assignment for the boxes two weeks
ago in their plain public view, the past will not be changed
by anything you do. Moreover, since the Alien is always
correct, from the point of view of the audience you do
not have free will.
But from your point of view---which should be regarded
as on an equal footing with theirs, at least operationally,
For if you try on some days to take both dinner and dessert,
then you cannot avoide the unmistakeable feeling that you can
*control* whether or not the main box contains dinner. As
a functioning organism you must adopt this point of view
that you *can* decide. Philosophical niceties such as "oh, well,
it's all determined what you will do" are not usefully
descriptive of your actions or your situation.
If you were part of a team who every night had to make
a *decision* as to whether to take one or both boxes,
your patter about not having choice, or about the contents
of the boxes being already determined, would be received
with jeers and sneers by the others. You would quickly
abandon the language-modality  of determinism, and adopt
the language-modality of free will.
Many people here are quick to disparage the concept of
free will as utter nonsense. But we compatibalists counter
by emphasizing its utility in daily discourse, and are wont
to remind our critics of the primary role of language in
describing not only our world, but necessarily a world
in which we devices are embedded.
As in every night when you hold arguments with the fools
who want to take both boxes, you (and they) really are
agreeing that choice is possible.
Go ahead if you want and discard the notion of free will,
but are you going to also discard the notion of a machine
(e.g. you) being able to make a decision? I have not
heard any of the anti-compatiblists answer this question.
 Language-modality is my much more sophisticated,
refined, and scientific term for what John Clark calls
More information about the extropy-chat