From: Harvey Newstrom (mail@HarveyNewstrom.com)
Date: Tue Jun 03 2003 - 20:21:30 MDT
Robin Hanson wrote,
> >In the case of the Doomsday Argument, which purports to do statistical
> >analysis on all populations past and future, we would need to pull random
> >samples from all time periods including past and future.
>
> No law says that statistical analyses require random samples.
Agreed. But I think DA implies random sampling. Part of the argument seems
to be, "What are the chances that a random sampling of all possible people
would choose me this early in our development?" It implies that random
sampling had occurred from all future possibilities and then tries to
explain why such an early sample was obtained. If we concede that our
current sampling methods can't reach into the future and are limited to the
current time or earlier, then there is no explanation needed for why future
samples did not result.
> The argument that data from the past can never support inferences about
> the future would prevent us from ever drawing conclusions about
> the future.
To clarify, I think that data from the past does not represent future
states. Extrapolating trends from the past definitely can predict future
states. As such, our ever-expanding civilization predicts that we will grow
larger and older indefinitely into the future. This is the opposite
conclusion than DA reaches. The reason is that DA is not using past data to
predict future trends. It tries to use past data samples to represent
future data sampling, which fails.
-- Harvey Newstrom, CISSP, IAM, GSEC, IBMCP <www.HarveyNewstrom.com> <www.Newstaff.com>
This archive was generated by hypermail 2.1.5 : Tue Jun 03 2003 - 20:33:29 MDT