3 cheers! ....and a history question.
Bill Taylor
ftp backgammon server.
GREAT NEWS !!
3 hearty cheers to Andreas for this effort. I'm sure we are all greatly in your
debt (or soon will be). You will score points in heaven for this Andreas!
-----------
The history question.
Some while ago, it was announced that a computer had beaten the then world
backgammon champion. I believe the alleged "match" took place straight after
the champs had been held.
This result was widely touted for some time by AI people as a marvel, and the
first time a machine had beaten the best human at some game.
However, I also dimly recall reading that it was barely a match at all; in that
it was only about 7 (?) games long. Hardly decisive, I'm sure you'll agree!
So anyway; can someone here shed light on all this. For instance...
1.When was it.
2.Who was the world champ involved.
3.Who/what were the machine/program/programmers.
4.What really happened.
5.Have any similar matchings occurred since.
And while I'm at it,
6.Just how good ARE machines at backgammon now? I know at chess they've
become close to grandmaster level. But at go, for instance, they're still a
biiiiiig dissappointment (brute-force is ineffective here).
So what about backgammon.
Are there ANY standard games where machines are unquestionably the best ?
Othello??
---------------------------------------------------------------------
Bill Taylor w...@math.canterbury.ac.nz
---------------------------------------------------------------------
Humans - God's attempt to pass the Turing test.
---------------------------------------------------------------------
Larry Hunter
Bill Taylor asks:
Some while ago, it was announced that a computer had beaten the then world
backgammon champion. I believe the alleged "match" took place straight after
the champs had been held.
1.When was it.
2.Who was the world champ involved.
3.Who/what were the machine/program/programmers.
4.What really happened.
5.Have any similar matchings occurred since.
6.Just how good ARE machines at backgammon now?
I don't have the answers directly, but the reference on that match is:
Berliner, H.J. "Backgammon Program beats world champion (Performance Note),"
AI Journal, 14 (1980) pp. 205-220.
Somebody borrowed my copy and never returned it, so I can't check the answers
directly. As for the current state of the art, you might try contacting Hans
Berliner, probably the foremost researcher in the computer gameplaying
community. He is a CS prof at Carnegie Mellon, and can be reached at:
Please post a summary of the responses you get!
Larry
--
Lawrence Hunter, PhD.
National Library of Medicine
Bldg. 38A, MS-54
Bethesda. MD 20894
(301) 496-9300
(301) 496-0673 (fax)
hun...@nlm.nih.gov (internet)
Tim Weaver
>where the computer had a 3 and 5 point holding backgame against a five
>point prime and with weak timing, but was down about 45 pips. It rolled
>2-2 and evaluated the best chance of winning as going into a race, so it
>came up 22/24(2) and two other 2's. Then it rolled boxes twice, won the
>race, and won the match. Barclay Cooke wrote the article.
>
>(and so goes another stroll down memory lane....)
>
Well, 3 consecutive doubles is tough to beat in any end game (let alone
2 consecutive boxes). Since there is an element of luck in backgammon,
it doesn't suprise me that a computer may beat a grand master. It's
likely to happen a great deal sooner and more often than in chess.
I have one question, was the computer rolling the dice or were they
being rolled and punched in? I play on a game that rolls an amazing
number of consecutive doubles for one player or the other and sometimes
it's just straight consecutive. Results on an end game defy all
strategy.
--
+-----------------------------+--------------------------------------------+
|Timothy E. Weaver | Kalamazoo College | (616) 383-5656 |
|Database Programmer/Analyst | 1200 Academy | These are MY opinions!|
|email: twe...@kzoo.edu | Kalamazoo MI 49006 | Mine!! Mine!! Mine!!|
Shuman Lloyd Lee
exactly the same.... somebody help me out?
It was in the late 70s or early 80s when a b'gammon program (I think
Hans Berliner was involved in its development) defeated the winner of
the world open (or was it the winner of the world amateur? David
Leibowitz?) in a 5 point match.
I remember this because there was an article in Les Boyd's publication
where the computer had a 3 and 5 point holding backgame against a five
point prime and with weak timing, but was down about 45 pips. It rolled
2-2 and evaluated the best chance of winning as going into a race, so it
came up 22/24(2) and two other 2's. Then it rolled boxes twice, won the
race, and won the match. Barclay Cooke wrote the article.
(and so goes another stroll down memory lane....)
in reply to:
Excerpts from netnews.rec.games.backgammon: 25-Jun-92 3 cheers! ....and
a history.. by Bill Tay...@math.canterb
Per Starb{ck
> It was in the late 70s or early 80s when a b'gammon program (I think
> Hans Berliner was involved in its development) defeated the winner of
> the world open (or was it the winner of the world amateur? David
> Leibowitz?) in a 5 point match.
If we are talking about the same match it was a 7 point match and
another opponent. From a local Twenex system (Yes, there are still
some of those running!):
> @xbkg
> Good morning
> This is C-MU's BKG V9.9, Alias Mighty-B,
> 7-1 Conqueror of World Champion Luigi Villa.
> Are you playing White, Black, Spectator, or a Friend?
> W, B, S or F:
Hans Berliner wrote an article about Mighty-B in Scientific American,
June 1980. Or so I've heard anyway---I haven't seen or read it
myself.
BTW: Bg is one of the games played at the Computer Olympiad, played in
London every year since 1989. (The 4th Computer Olympiad will take
place this summer August 5--11.) What bg programs have won earlier
years? Has someone here participated?
-- "
Per Starback, email: star...@student.docs.uu.se, irc: starback
"Life is but a gamble! Let flipism chart your ramble!"
Andreas Schneider
>Andreas Schneider <mar...@ouzo.rog.rwth-aachen.de> announces an impending
>ftp backgammon server.
The server will have nothing to do with ftp, but will work with telnet, but
who cares.
>3 hearty cheers to Andreas for this effort. I'm sure we are all greatly in your
>debt (or soon will be). You will score points in heaven for this Andreas!
Great! The first ones :-)
>The history question.
>
>Some while ago, it was announced that a computer had beaten the then world
>backgammon champion. I believe the alleged "match" took place straight after
>the champs had been held.
>
>However, I also dimly recall reading that it was barely a match at all; in that
>it was only about 7 (?) games long. Hardly decisive, I'm sure you'll agree!
Actually it was only 5 games, but the program won 7-1 !!
>So anyway; can someone here shed light on all this. For instance...
>
>1.When was it.
It was in June 1979 (yes! THAT long ago!)
>2.Who was the world champ involved.
It was Luigi Villa from Italy
>3.Who/what were the machine/program/programmers.
Hans Berliner, an excellent chess player (world correspondence-chess
champion). He was supported by the famous Paul Magriel, who helped
refining some algorithms and improving the program's overall performance.
>4.What really happened.
Villa got backgammon world champion 1979 in Monte Carlo and played against
Berliners BKG 9.8 program after that.
The scores were the following:
game BKG 9.8 Villa
1 2 - 0
2 1 - 0
3 2 - 0
4 0 - 1
5 2 - 0
Berliner won $5000 and wrote afterwards:
Villa, who only a day earlier had reached the summit of his backgammon
career in winning the world title, was disconsolate. I told him that I
was sorry it had happened and that we both knew he was really the
better player.
Several weeks later I analyzed the games in some detail [...]. There was
no doubt that BKG 9.8 played well, but down the line Villa played better.
He made the technically correct plays almost all the time, whereas the
program did not make the best play in eight out of 73 nonforced situations.
Only one of the mistakes, however gave the program any trouble. An expert
would not have made most of the errors, but the could be exploited only a
small percent of the time.
That's wonderful in backgammon: Even a computer program can win against
the world champion, although it plays worse :-)
>5.Have any similar matchings occurred since.
I never heard about a world champion who dared playing against a computer
in the public after Villa :-)
>And while I'm at it,
>
>6.Just how good ARE machines at backgammon now? I know at chess they've
>become close to grandmaster level. But at go, for instance, they're still a
>biiiiiig dissappointment (brute-force is ineffective here).
>
>So what about backgammon.
All backgammon programs I played are miserable (and I'm not a champion at
all). BKG 9.8 was never available on the market as far as I know and Berliner
hasn't publshed anything about it since his article (it was in the Scientific
American if I recall it right - Don't ask me for the year). If he did: please
let me know! In his article he also described how his program worked. There
are really some very nice ideas in the program.
>Are there ANY standard games where machines are unquestionably the best ?
>Othello??
I heard that checkers has been solved completely and humans can't win against
the best programs anymore.
>---------------------------------------------------------------------
> Bill Taylor w...@math.canterbury.ac.nz
>---------------------------------------------------------------------
> Humans - God's attempt to pass the Turing test.
>---------------------------------------------------------------------
(but did she succeed?)
Andreas
--
Email: mar...@ouzo.rog.rwth-aachen.de
mar...@pool.informatik.rwth-aachen.de
Mail: Andreas Schneider, Kapuzinergraben 8, 5100 Aachen, Germany
Phone: +49-241-35562
David Escoffery
In article <marvin.709513651@rama> mar...@rama.informatik.rwth-aachen.de (Andreas Schneider) writes:
| I never heard about a world champion who dared playing against a computer
| in the public after Villa :-)
The March-April (Volume 2, Number 2) and May-June (Volume 2, Number 3)
issues of "Inside Backgammon" featured a two-part article by twice world
champion Bill Robertie reporting on the state of backgammon computers and
annotates his match against Dr. Gerald Tesauro's program, TD-Gammon. and I
trust that Kent will not mind me reproducing some of Robertie's article for
the benefit of those on the net.
Anyways, Dr. Gerald Tesauro is employed at the IBM Research Labs in White
Plains, New York. His latest backgammon program is called TD-Gammon and is
based on neural network theory. TD-Gammon runs on an RS-6000 workstation. In
the beginning, it was programmed only with the actual rules of backgammon and
the ability to generate legal moves. No knowledge was programmed into it as to
what constituted a good or a bad move. It "knew" nothing about making points,
hitting blots, or anything else. Over the course of several months, it played
300,000 games against itself. In the beginning, it picked its moves at random
from the list of possible legal moves. After each game, a logic routine made
informed guesses as to what moves in the previous game may or may not have
been errors, based upon a sophisticated mathematical theory of learning. The
program's positional evaluator routine continuously modified itself, based
upon its results in previous games.
The article and the games are interesting, as Robertie annotates and analyzes
each move. At the end he summarizes ..
"TD-Gammon and I played most of a day, a total of 31 games. I won 19 points,
an average of 0.61 points per game. On balance, I was lucky. Game 16, as you
saw, could easily have gone the other way at the end, a 16-point swing by
itself [Robertie had to throw a double on his last roll to win]. My estimate,
after reviewing the entire session, was that I would do well to average 0.20
to 0.25 points per game. This figure makes TD-Gammon the strongest backgammon
program in existence, most likely better than Berliner's program of 13 years
ago, although that's no longer available for comparison.
"Not only is TD-Gammon interesting as a backgammon program, it represents an
astonishing achievement for the neural network approach to artificial
intelligence. Remember that this program has no human knowledge built into it.
Everything it "knows", it deduced by playing against itself, then improving by
applying sophisticated mathematical learning algorithms to the results of its
games.
"Just before going to press, we received word that Malcolm Davis and Paul
Magriel made journeys up to White Plains to match wits with TD-Gammon. Malcolm
Davis broke even in 12 games, although TD-Gammon won 8 out of 12. Paul Magriel
got backgammoned while playing a favorable back game and ended up negative for
the session."
There was one anecdote Robertie relates that I found interesting. In its
300,000 games of experience, Robertie felt that TD-Gammon has not "learned" to
slot the 5-point with an opening 4-1, regarding the split on an opening 21,
41, or 51 as superior to the slot. After rolling out the opening position
1000 times, the program finds that while 13/9 24/23 makes it exactly even
money, slotting the 5-point leaves it an underdog by 0.05 points.
===========================================================================
David Escoffery Tel: (408) 427-7718
The Santa Cruz Operation Internet: dav...@sco.COM
P.O. Box 1900
Santa Cruz, CA 95061
===========================================================================