On Thu, 18 Feb 1999 blstuart(a)bellsouth.net wrote:
In message
<01be5b3b$65657aa0$d3c962cf@devlaptop>, "Mike" writes:
Speaking of _Adikesence of P1_ .... Does anyone
know the SciAm issue date
for the referenced matchbook neural net ???
It's been a long time since I read the book (I really should
read it again, now that the subject has come up). But if
he're refering to the article I think, it's by Martin
Gardner. Unfortunately, I can't find a full reference.
(Somewhere I've got a photocopy of it, but I can't find it
right now.) Anyway, I do find that it came from a 1962 issue.
In the article, Gardner defines (I can't remember if he references
anyone else) a simple game played with six pawns played on a 3x3
board. Using 24 matchboxes, he shows a learning machine called
Hexapawn Educational Robot (HER).
I've got a fond spot in my heart for this one of Gardner's
articles. A simplified version was printed under the title
"How to Play Hexapawn" in the collection called *The New Reader's
Digest Treasury for Young Readers,* a copy of which was given to
me by a relative when I was a kid. It sparked my interest in AI
which reached it climax about seven years ago when I defended my
dissertation in machine learning.
Hexapawn as a BASIC program was in David Ahl's first (DEC) release
of "BASIC Computer Games". The original Scientific American article
inspired the story "Without a Thought" by Fred Saberhagen, the first
of his "Berserker" short stories about robots left by an ancient
civilization ripped off without credit for the episode of the old
Star Trek episode "The Doomsday Machine". Hexapawn is a fun game,
even though you know damned well you'll lose later. (Like arcade
games in a way -- however well you do at Space Invaders, they'll
just keep crapping on you until you die only faster).
--
Ward Griffiths
"the timid die just like the daring; and if you don't take the plunge then
you'll just take the fall" Michael Longcor