From: James Rogers (jamesr@best.com)
Date: Wed Aug 20 2003 - 22:49:25 MDT
On 8/20/03 4:37 PM, "Samantha Atkins" <samantha@objectent.com> wrote:
> Isn't this another way of reiterating your fundamental assumption without
> actually supporting it? I have no way of accurately guessing how many
> equally efficient workable AGI designs there might be at this point.
Almost everyone who responded to my assertion did not grok it. That may
have been a communication failure on my part or simply that I assumed the
audience had sufficient contextual background to figure it out. Mixed with
a bit of apparent reading non-comprehension. I've been meaning to get back
to this but stepping through it from the primitives will be painfully
lengthy so time is an issue. The reasoning behind my assertion is actually
fairly elegant in a formal sense and you can prove it with the clever
application of things we've already discussed on this list; I thought
someone might put the pieces together.
So not tonight, as I have more important things pressing at the moment. I'm
analyzing the output of a very clever piece of software as it analyzes
Plasmodium genomes this evening. Speaking of which (and admittedly
off-topic), I developed a lovely new algorithm for doing local alignments of
sequences that appears to blow the pants off of the standard Smith-Waterman
family of algorithms both in terms of time AND space by adapting some of the
core algorithms I developed for the AI work. The applicability of this
stuff seems endless even without AI per se.
Like the Novamente people, we are finding a rather immediate opportunity for
these technologies in biotech. I'm *finally* spending a huge chunk of my
time on this, which has been helpful in hitting milestones at a brisk pace.
Off to work I go,
-James Rogers
jamesr@best.com
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:42 MDT