COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\|\\M1BASL30\M2NGR40L\M3NGR25\M4NGR20\F2\CSTANFORD UNIVERSITY
C00010 ENDMK
C⊗;
\|\\M1BASL30;\M2NGR40L;\M3NGR25;\M4NGR20;\F2\CSTANFORD UNIVERSITY
\F3\CSTANFORD, CALIFORNIA 94305
\F4ARTIFICIAL INTELLIGENCE LABORATORY\←L\-R\/'7;\+R\→.\→S   Telephone:
\←S\→.415-497-3330

\F1\COctober 7 1977


Mr. Ken Hanson
317 San Vincente Ave.
Salinas, Calif. 93901

Dear Ken:

\JI am afraid that I let the ball drop by not replying to
your last letter.  I was temporarily diverted from working on the checker
program and then I went away for a rather long trip.  It also seemed to me
that you did not sound too anxious to get re-involved with my
program, not that I can blame you.  I am now back home and ready to
do some more work and I am assuming that you are also back from your travels.

At the moment, I am involved in quite a different checker program,
trying to put a cut down version onto a micro-computer.  You
probably have seen the little chess machine that sells for $200
and also the game attachments that hook into a TV
set and that allow one to play certain rather trivial games.
One of the manufactures of one such machine got the idea of adding
checkers to the list of things that his machine would do and they have
asked me to help.  It looks like it will be possible to do this although
all sorts of corners will have to be cut to get something that will
go into the limited memory space and that will not take forever to make a
move.  Anyway, it is lots of fun trying.
 
I will not be able to store many moves, and certainly I will not be able
to including a learning routine.  Although the gadget will be only of interest
to rank amateurs, I do want it to start off on the right foot and so I propose
to store most of the acceptable answers to all possible first moves and also
most of the acceptable counter replies to all of the 7 different possible
replies to each of the 7 possible first moves.  Of course,  I have the
acceptable moves from Lee's with your corrections but I do not have counter
replies to obviously bad second moves.  For amateurs I will need these.
Could I impose on you for this information?

I propose to have the program choose moves from these lists by, in effect,
flipping a weighted coin so I would like to have the moves rated
as to their relative values.  I can then have the machine chose the better moves
more often than the others and still not have it play the same every time.

While I am writing, I would like to have your reaction to the other projects
that I suggested in my last letter, and to which you did not comment.  In
case you did not keep this letter I will quote the paragraphs in question.
 
"As a second project, I would like to see your work in debugging the master
play made generally available, perhaps by publishing a revised and
anotated Lee's Guide.  We probably could not call it a Lee's Guide, for
copyright reasons, and this might be one reason for extending it to
include data from Kear, Bolan and even perhaps Tescheleit as well.
Actually, I think that it could very well be much more than just another
checker book.  With the computer tools that we now have available, it
should be rather easy to produce the text, suitable for photo-ofset
printing, for a book with information presented in two ways, firstly, in
the usual game with variations form and secondly, with a listing of board
positions, with the recommended move or moves and perhaps even
cross-referenced to the games in which these positions occur.  If we work
out a suitable format then perhaps the ACF might even want to back the
project and we could then get help in the form of listings of other games
that could be included, and it could well be made into an authoritative book.

I then have an even more ambitious project in mind, in that I would like
to have additional information appended to the book listings, not as I
earlier thought, as to why the chosen move is the best one but rather why
each of the other possible moves is not a good one. In this way one gets a
great deal more information out of each move and this could be used by the
program to improve its play.  This would take a lot of work and it could only
be done by someone with a lot of checker knowledge, such as yourself."

I judge that you did not think too much of these ideas but I would still like
to know why not.\.

\←L\→S\←R\-L\/'2;\+L\→L

Sincerely,



Arthur L. Samuel

\←S\→L
ALS:pdp10