Saibal Mitra wrote:

----- Original Message -----
From: "Quentin Anciaux" <[EMAIL PROTECTED]>
To: <everything-list@eskimo.com>
Sent: Monday, June 20, 2005 11:37 PM
Subject: Measure, Doomsday argument


> Hi everyone,
>
> I have some questions about measure...
>
> As I understand the DA, it is based on conditionnal probabilities. To
somehow
> calculate the "chance" on doom soon or doom late. An observer should
reason
> as if he is a random observer from the "class" of observer.
>
> The conditionnal probabilities come from the fact, that the observer find
that
> he is the sixty billions and something observer to be "born". Discover
this
> fact, this increase the probability of doom soon. The probability is
> increased because if doom late is the case, the probability to find myself
in
> a universe where billions of billions of observer are present is greater
but
> I know that I'm the sixty billions and something observer.


This is a false argument see here:

http://arxiv.org/abs/gr-qc/0009081


To calculate the conditional probability given the birthrank you have you
must use Bayes' theorem. You then have to take into account the a priori
probability for a given birthrank. If you could have been anyone of all the
people that will ever live, then you must include this informaton in the
a-priori probability, and as a result of that the Doomsday Paradox is
canceled.

I don't think the cancellation argument in that paper works, unless you already *know* the final measure of one type of civilization vs. another from the perspective of the multiverse as a whole. For example, if I know for sure that 50% of civilizations end after 200 billion people have been born while 50% end after 200 trillion have been born, then it's true that observing my current birthrank to be the 100 billionth person born, I should not expect my civilization is any more likely to end soon, since 50% of all observers who find themselves to have the same birthrank are part of 200-billion-person civilizations and 50% of all observers who find themselves to have the same birthrank are part of 200-trillion person civilizations. But if I don't know for sure what the measure of different civilizations is, suppose I am considering two alternate hypotheses: one which says 50% of all civilizations end after 200 billion people and 50% end after 200 trillion, vs. a second hypothesis which says 99% of all civilizations end after 200 billion people and 1% end after 200 trillion. In that case, observing myself to have a birthrank of 100 million should lead me, by Bayesian reasoning, to increase my subjective estimate that the 99/1 hypothesis is correct, and decrease my subjective estimate that the 50/50 hypothesis is correct.

Jesse


Reply via email to