From: Ramez Naam (mez@apexnano.com)
Date: Tue Apr 29 2003 - 23:06:12 MDT
From: gts [mailto:gts_2000@yahoo.com]
> Ramez Naam wrote:
> > So what is there to support the Self Sampling Assumption?
>
> Let's start with this example (I'm making this example up as
> I write):
>
> You and 99 other people are standing shoulder to shoulder in
> a line. Someone from behind the line places green or yellow
> hats on the heads of everyone in the line, including you. You
> and everyone else in the line are now sporting a hat, but
> none of you know which color of hat you have, and none of you
> know the colors of the hats of others.
>
> All you know is that 90% of the hats are green.
>
> Now someone asks you to guess the color of your own hat.
>
> Would you guess green? Of course you would. You would reason
> that given that you are a random sample of the 100 people in
> the line, and that given that 90% of the hats are green, you
> are very likely to be wearing a green hat. You made this
> guess using the Self-Sampling Assumption (SSA). And you were
> right to make that guess.
This is an interesting example, but can you think of one more
realistic?
As far as this example goes, I would guess that I was wearing a green
hat if I had no other information whatsoever, but I would recognize
that it was an extremely weak guess.
Why? Because the distribution of hats is probably not random. It
depends on the behavior of the person handing them out. That person
may be giving yellow hats to all the bald men, and green hats to
everyone else, or using any other criteria. If I knew that the person
handing out the hats was really doing so randomly, then green would be
a stronger guess. But until I know what the distribution pattern is,
all of my guesses are weak.
The same problem holds for the Doomsday Argument. Whether or not a
species goes extinct is not random. It depends upon qualities of that
species. So any guess as to humanity's future based on an assumption
that a human Doomsday is random is going to be a very weak guess. It
comes back to what Rafal and Harvey said in the other branch of the
thread - we just don't have enough information for DA to be useful.
cheers,
mez
This archive was generated by hypermail 2.1.5 : Tue Apr 29 2003 - 23:16:27 MDT