Man vs Computer Checkers Championship
Alex Lopez-Ortiz
Today's newspaper has a picture of Marion Tinsley palying a computer with
the caption [my comments are in brackets].
World checkers champ Marion Tinsley ponders his next move against
the Chinook Supercomputer in London, England, yesterday during the
World Draughts [checkers] Championship. The computer is ranked
second in the world [the checkers program was written by J.
Schaeffer of the Univiersity of Alberta]. Tinsbury, of Tallahasee,
Florida, as of yesterday had won one game and drawn five in a 40-game
series against the computer.
Does anybody has more information on the match?
Alex
--
Alex Lopez-Ortiz alop...@maytag.UWaterloo.ca
Deparment of Computer Science University of Waterloo
Waterloo, Ontario Canada
Darse Billings
>Today's newspaper has a picture of Marion Tinsley palying a computer with
>the caption [my comments are in brackets].
> World checkers champ Marion Tinsley ponders his next move against
> the Chinook Supercomputer in London, England, yesterday during the
> World Draughts [checkers] Championship. The computer is ranked
> second in the world [the checkers program was written by J.
> Schaeffer of the Univiersity of Alberta]. Tinsbury, of Tallahassee,
> Florida, as of yesterday had won one game and drawn five in a 40-game
> series against the computer.
>Does anybody has more information on the match?
We here at the U of Alberta are receiving daily updates of the match.
I will speak to the sources and see if they will provide this
information to rec.games.chess (they are posting it locally already).
(Incidentally, if this doesn't seem to be the appropriate newsgroup,
then be sure to vote 'yes' to the creation of rec.games.strategic).
Failing all else, I will post updates every few days (plus any day which
has a decisive result). Here is some background information and the
current situation:
Dr. Marion Tinsley is a professor of pure mathematics from Tallahassee,
Florida. He has been the reigning world champion of 8x8 checkers for
more than 40 years, and in that time he has lost five games (some by the
touch move rule). Some consider his play to be close to perfection.
But he is human. Chinook is not. The program was written by a group at
the University of Alberta, headed by Jonathan Schaeffer who is a renown
programmer of computer chess (Phoenix) and an authority on Alpha-Beta
search and game tree search.
Chinook exploded onto the world checker scene a few years ago, quickly
becoming the second highest rated checker player in the world, behind
only Tinsley. It earned the right to challenge Tinsley for the world
championship title, but the match was delayed for various reasons.
As I understand it, the international federation governing checkers
later forbid Tinsley to play a computer for the title. This prompted
Tinsley (who is a very sporting man) to resign his title, and a match
was arranged for the "de facto" world championship.
That match began Monday, August 17, 1992 in London England. The match
is scheduled for 40 games, to be completed by August 29. The prize fund
is not less than $20,000 -- but I am not sure of the exact amount.
The day by day results (sorry, I do not know the color assignments):
Monday August 17:
Game 1: Chinook 1/2 Tinsley 1/2
Game 2: Chinook 1/2 Tinsley 1/2
Tuesday August 18:
Game 3: Chinook 1/2 Tinsley 1/2
Game 4: Chinook 1/2 Tinsley 1/2
Game 5: Chinook 0 Tinsley 1
Game 6: Chinook 1/2 Tinsley 1/2
Wednesday August 19:
Game 7: Chinook 1/2 Tinsley 1/2
Game 8: Chinook 1 Tinsley 0 (!)
Thursday August 20:
Game 9: Chinook 1/2 Tinsley 1/2
Game 10: Chinook 1/2 Tinsley 1/2
Game 11: Chinook 1/2 Tinsley 1/2
Game 12: Chinook 1/2 Tinsley 1/2
After 12 games the match stands:
Chinook 6 pts.
Tinsley 6 pts.
Wednesday's loss by Tinsley was his first in more than 10 years, and
sixth since he attained the world championship title. The games have
reportedly been very long (on the order of five hours a game), and
Chinook has been playing with its typical computer style of complicated
games.
This may eventually wear down Tinsley, and give Chinook a better chance
of winning the match. Most experts still consider Tinsley to be the
better player, but he is not a young man.
If Tinsley is to win this match, he must capitalize on Chinook's
weakness in the opening. In checkers, unlike chess, the opening is of
critical importance, since the result of a bad opening is often a lost
position, rather than just an inferior or difficult position. Chinook
generated its own opening book, but is still vulnerable to known traps,
which have accounted for the majority of its losses.
On the other hand, Chinook has recently been parallelized to increase
performance, and its omniscient endgame database now includes all
positions with seven or fewer pieces and many eight piece positions.
Perhaps it has finally overtaken the human master in playing strength.
As a disclaimer, I should mention that while Jonathan Schaeffer is my
supervisor, I am not directly involved in the Chinook project, and some
of the information I have provided here may not be perfect. I can post
more details on the Chinook program if there is sufficient interest.
--
- Darse Billings, 2100 CM, 7 kyu.
Go is better than Chess. Poker is more lucrative. Sex is more fun.
Robert Hyatt
>
>
>Today's newspaper has a picture of Marion Tinsley palying a computer with
>the caption [my comments are in brackets].
>
> World checkers champ Marion Tinsley ponders his next move against
> the Chinook Supercomputer in London, England, yesterday during the
> World Draughts [checkers] Championship. The computer is ranked
> second in the world [the checkers program was written by J.
> Schaeffer of the Univiersity of Alberta]. Tinsbury, of Tallahasee,
> Florida, as of yesterday had won one game and drawn five in a 40-game
> series against the computer.
>
>
>Does anybody has more information on the match?
>
match. Tinsley has only lost 6 games total (including the one game
to Chinook) in the past 35 years or so.
Bob
--
!Robert Hyatt Computer and Information Sciences !
!hy...@cis.uab.edu University of Alabama at Birmingham !
Bruce_L...@transarc.com
It's incredible to me that Tinsley could maintain a pace of 12 games in 4 days.
Weekend chess tournaments used to be like this (3 games Saturday, 2 games
Sunday, at time controls which led to 5-hour games), but that format has been
almost universally abandoned because it was too strenuous. International
chess tournaments are held at 1 game a day, and even this is too strenuous
for many players.
Eric S. Perlman
It all depends on 1) The number of moves per game and 2) the time
control. Scrabble tournaments, for example, are played with a G/25 time
control, and usually run 5-8 rounds per day. On the other hand, the
average tournament scrabble game lasts around 15 moves, not 45 or so.
--
"How sad to see/A model of decorum and tranquillity/become like any other sport
A battleground for rival ideologies to slug it out with glee." -Tim Rice,"Chess"
Eric S. Perlman <per...@qso.colorado.edu>
Center for Astrophysics and Space Astronomy, University of Colorado, Boulder
Kathy Watts
>
>>Does anybody has more information on the match?
>
>We here at the U of Alberta are receiving daily updates of the match.
>
>I will speak to the sources and see if they will provide this
>information to rec.games.chess (they are posting it locally already).
>(Incidentally, if this doesn't seem to be the appropriate newsgroup,
>then be sure to vote 'yes' to the creation of rec.games.strategic).
>
>
>Go is better than Chess. Poker is more lucrative. Sex is more fun.
Darse, thanks for the info, but please realize that this is a one-time
occurance. There will likely not be another match like this in checkers again.
This group ir not rec.games.checkers, but we are very interested because of the
obvious similiarities to the more difficult questions of computers and chess.
Please don't go using this as an excuse to over-generalize the match into the
basis for an unneeded (IMHO) new newsgroup.
I myself don't find the traffic in this newsgroup overwhelming, and enjoy the
number of subjects that come up. And it's easy enough to skip things that
don't interest me.
If any campaign is needed, it's one aimed at getting clear descriptions into
the subject lines (short, sweet, and to the point; and with a clear indication
of whether or not a new subject is being introduced on the heels of a dying
one.
-Hal Bogner
Janet Service
They are playing 20 moves per hour. The games on Monday lasted approx.
4 and 5 hours, so only 2 games were played on that day, therefore the
other 2 games were played on Wednesday (a scheduled rest day) thus
keeping to the agreed 4 games per day schedule.
Steven L Harrington
This is really ridiculous! This is effectively giving the computer a huge
practical advantage in the match, where fatigue in the human is inevitable
(age notwithstanding). I have no idea why Tinsley would agree to such
match conditions, other than his oft-cited sportsmanship; however, he
has really carried his sportsmanship to an extreme to willingly give
his opponent such an edge. If he were playing a human I could fathom
these conditions as _both_ participants would be subject to fatigue,
but against a machine it's ludicrous. I confess that I am a biased
observer, but it seems that many human-computer matches are played
on a field that favours the machines (i.e. Harvard Cup @ ~20mins per game
per side - a time control that clearly favours the computers). I just
hope that Tinsley doesn't live to regret such incredible match conditions...
It is my hope that the best player win the match, not the one with the
best stamina (infinite stamina).
-steve harrington
-texas a&m university
Eric Schiller
but I have been to busy and don't have easy access to the info.
It is a one-off event and won't waste much bandwith anyway.
Ray Keene thanks you for posting!
Eric Schiller
Kathy Watts
match would become an endurance contest, rather than a test to see whether he
or Chinook played checkers better. But if Chinook wins on endurance, Schaeffer
will have no victory at all...better to slow the match down a bit and keep
focused on its purpose.
I hope that this suggestion is acted upon.
In Albuquerque last November, the panel discussion on the future of computer
chess brought out some interesting statements by Murray Campbell on behalf of
the Deep Thought team. While some panelists were consumed by scientific or
artistic goals, DT is being optimized to try to defeat GMs. So Murray
explained that one of their tasks is to analyze GM games and find the types of
positions in which DT could get them to go wrong - rich, complex tactical
position. DT, of course, may be able to out-calculate GMs in these positions
often enough to win games - winning matches will depend upon bringing these
positions about frequently.
Murray was challenged by another panelist for expressing an idea which was the
very opposite of the goal of chess as an art: simplicity, or elegance. To
which he replied, "This isn't about art, it's about winning!" (Forgive me if
I've lost something in my paraphrase - it's been 10 months.)
I expect to see a paper in the ICCA journal entitled "Chess as Sport" or,
if Chinook simply plays Tinsley to exhaustion, "Checkers as Sport." Although,
in fairness, there is a big difference between the DT strategy outlined above
and simply tiring out the opponent by denying adequate rest to the human
between games.
My point is really that winning has replaced solving as the goal. A long time
ago, winning looked a way of measuring progress toward solving chess with a
machine; now that winning against the world champion seems possible, we need to
be clear on whether the initial goal is still being pursued.
Hal Bogner
Paul Rubin
that the match would become an endurance contest, rather than a
test to see whether he or Chinook played checkers better. But if
Chinook wins on endurance, Schaeffer will have no victory at
all...better to slow the match down a bit and keep focused on its
purpose.
I hope that this suggestion is acted upon.
Tinsley has been champion for several decades and has defended the title
many times, probably in similar matches. I can't believe he didn't
know what he was getting into before this match started.
What bothers me more is how some of the longer games eliminated a
rest day. That kind of thing could never happen in a world chess
championship match. In chess championships, in addition to scheduled
rest days, each player gets a certain number of personal rest days
which he can use whenever he wants.