From: justin corwin (outlawpoet@gmail.com)
Date: Tue May 24 2005 - 17:59:25 MDT
On 5/24/05, Eliezer S. Yudkowsky <sentience@pobox.com> wrote:
> Peter C. McCluskey wrote:
> > Eric Baum
> > makes some moderately strong arguments in his book What is Thought?
> > against your claim. If your plans depend on a hard takeoff and your
> > reasons for expecting a hard takeoff are no better than the ones I've
> > run across, then I'm pretty certain you will fail.
>
> Eric Baum calculates 10^36 operations to get intelligence, based on the number
> of organisms that have ever lived. To see why this number is wrong you may
> consult http://dspace.dial.pipex.com/jcollie/sle/ or for more information
> George Williams's "Adaptation and Natural Selection."
Er, quite aside from what value we may give the amount of 'work'
evolution went through designing a mind, it seems patently absurd to
imagine that even a corrected figure would have much to do with the
difficulty of designing something.
Design is not axiomatic computation. It is not required that you
compute the phase space that contains the machine that you wish to
build, nor that you even have a model that entirely contains all the
relevant complexity. It only requires that your theory narrows the
design space to the point where you can be sufficiently lucky to do it
right. Better theories, and luckier people may change this value
significantly.
If this is a nervous kind of line of reasoning for some people, I
think that's a good thing. It's important to recognize that most
'designs' are not actually for exact, specified things, they are
classes of possible things.
Of course, better designs are smaller spaces, and so finding optimal
designs is computationally expensive, depending on how lucky you are
to begin with, and what value you're using for optimal, and your
generation process, and a host of other things that makes it difficult
to estimate.
Baum's number is interesting, if bizarre. a quick ride on IDLE shows
that you would have to run a human-level computer (10^17(Moravec))
about 317097919837years to do it. Our computers today, are of course
not even worth talking about. I suppose we'd better start thinking
early, and put as many people on the job as possible.
-- Justin Corwin http://www.adaptiveai.com http://outlawpoet.blogspot.com outlawpoet@hell.com
This archive was generated by hypermail 2.1.5 : Tue Feb 21 2006 - 04:22:56 MST