ditzel@ssc-vax.UUCP (07/11/83)
Re: Chess Life and Review
Does anyone know what is happening with Chess Life (& Review). Tried to
pick up a copy at a magazine store and they told me that they were
no longer distributing it. Is this true?
Re: Korchnoi versus Kasparov
A couple of months ago these two played and Kasparov came out on top
after emerging from a hideous position, if enough people want the game
I put it on net.chess. Is Kasparov really as good as people are saying?
If he is Korchnoi is in for a bad time.
Re: Net.chess
How about more activity on this net? I simply can't believe amoung all
the people out in netland there are no chessplayers!
cldnewman@utcsrgv.UUCP (Ken Newman) (07/14/83)
This is a slightly unusual topic for this group, but I couldn't think of a better group to put it in. Does anyone know what the relative performance levels are between Othello masters and the best computer programs? Is the situation similar to the current chess state of affairs (i.e. programs not truly at the master (int. master?) level, masters not having much trouble beating any program)? Also, is a computer program still the world champ in backgammon (it was a while ago)?
WILKINS@SRI-AI.ARPA@sri-unix.UUCP (07/15/83)
From: Wilkins <WILKINS@SRI-AI.ARPA> I keep hearing about a program being the world backgammon champion, and this is not true. Furthermore, the falsehood is perpetrated by people in AI who should know better which also annoys me. (Fortunately, the author of the program plays no part in such perpetration). Hans Berliner's program did beat the reigning world champion in a 7-point match. Now, suppose the St. Louis Cardinals play a one-inning exhibition against the baseball team of Alcorn State (3 outs for each team). The Cardinals happen to get 3 outs without scoring while Alcorn State's best hitter happens to catch one good for a home run. Do you now say that Alcorn State's baseball team is the World Champion? No way. Likewise, there is no way you can claim the program is world champion because it got better luck in a match that was only a small fraction of the length of any reasonable championship match. So please stop perpetrating this falsehood and try to stop others from perpetrating it also. I have corrected some of the top people in AI who have said similar things in public. Then there are books like "Science Observed" by Jeremy Bernstein, which go out of their way not only to perpetrate such falsities but actually add to them. The first n pages are about this backgammon match, with lots of stress put on the fact that it wasn't just a program but an actual robot that won (bogus). He then went on to tell how Berliner told the downhearted Villa, "we both know you're the better player", insinuating that Villa was being humored (we all know how smart the computer is). I know Dr. Berliner, and he knows Villa is the better player and I feel certain that he meant every word sincerely. I certainly did not read the rest of Bernstein's book after wading thru the crap in the first few pages. The program does not play anywhere near a world-class game. I have a ready proof in the form of an offer you are free to give on my behalf to anyone who claims otherwise. I'll let you bet on the program against me and I'll give you 7 to 5 odds. The offer applies to either 7-point matches or simply games for money. 7 to 5 odds would be overwhelming between two world-class players, yet I'm only an intermediate player, not even an expert. (P.S. If you take my offer you'll likely lose a lot of money, as I will want a high enough stake to pay for my time.) In search of truth and knowledge, David Wilkins -------
newman@utcsrgv.UUCP (Ken Newman) (07/17/83)
I submitted an article recently in which I commented about
the world champion in backgammon being a computer program a
while ago, and got mail from someone asking my source for
that statement. Well, my return mail got eaten ("host not
found"), and then I accidentally deleted the path of the
person who sent me mail. So I don't know who you are, and
I'm posting my reply here instead. One of those days...
My information regarding computer backgammon was quite correct.
In a fascinating article that was the cover article of the
June 1980 Scientific American, by Hans Berliner, a program is
described that beat the world champion (then) Luigi Villa
of Italy, by the score of 7-1 in a $5,000 winner-take-all
match in Monte Carlo. It was the first time a computer program
had beaten a world champion at any board or card game. The
program is called BKG 9.8, and I recall seeing an ad in Byte
for a CP/M version of the program(!). Actually I'm not sure
if this was part of an official world championship tournament;
I guess there is a difference between beating the world
champ in a match and becoming the official world champion.
Pretty impressive at least though.
Ken Newmantrt@rti.UUCP (07/17/83)
Re: computer Othello (TM Gabriel), computer backgammon. Since noone else has yet replied, I will make an attempt. OTHELLO The CMU Othello (generic near-equivalent is 'reversi') program is probably better than any human. I do not believe it has an official title, though. In 1979, the first man-machine Othello tournament was held, and the humans (world champ and runner up) finished 1, 2. *But they each lost one of their 6 games against a machine.* Kathy and Dan Sprachlen (Sargon, Boris, Chess Challenger, etc.) had the best computer program, and the CMU program did well too. (I forget the CMU authors' names -- check the 1982 SIGARTs.) A year or so later a revised CMU program went 8-0 (!!) in an all-computer event. The world champion, Jonathan (?) Cerf, was in attendance and said the CMU program was probably better than he was, and declined to play a game. So, something of an AI milestone was reached with Othello -- superiority over mankind in a fairly complex board game -- but the whole thing was pretty much a fizzle. (Computers were total winners in Kalah years ago, but Othello was a much bigger challenge.) BACKGAMMON In 1979, the CMU backgammon program developed by Hans Berliner et. al. defeated the new world backgammon champ 5-1 (in 4 games). The program was not and is not the world champ, it just beat the champ in the only such match ever played. Four games is far too few to judge the relative strengths of two good players, and some (even Berliner) have said that the program made noticeably worse moves but was lucky (e.g. plenty of double sixes). The CMU program is certainly good though, and the win is an interesting historical footnote. Unfortunately, backgammon has such a large component of luck that even a 'perfect' computer backgammon program might have trouble achieving the recognition that it deserved. I have Berliner's 1979 writeup of the match, including the game score. Let me know via mail if you want a copy. Tom Truscott (duke!trt, trt.duke@udel)
mclure%SRI-Unix@sri-unix.UUCP (07/18/83)
Not really about chess but... I played BKG 9.9 extensively but finally gave up after winning 11 7 pt matches to 3 7 pt matches. Its doubling leaves a lot to be desired and is definitely the main weakness of the program. Last I heard, Dr. Berliner was converting part of BKG to LISP so that it could answer questions about its actions in a manner similar to Winograd's SHRDLU. Also he mentioned that he was preparing an endgame book so that BKG could simply look up the proper move in situations where contact between the two sides is broken. I'm not sure if this is only for when everyone is in the home-board or when contact is broken. Stuart