home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Classic Games for OS/2 Warp
/
Classic Games for OS2 Warp (1995)(IBM).iso
/
tdgammon
/
tdgammon.hlp
(
.txt
)
< prev
Wrap
OS/2 Help File
|
1995-05-26
|
32KB
|
809 lines
ΓòÉΓòÉΓòÉ 1. General help ΓòÉΓòÉΓòÉ
Backgammon is a game of both skill and luck in which two player move their
"men" in an obstacle race around the board, trying to get their men off the
board before their opponent does.
OS/2 TD-Gammon lets you play against one of the strongest backgammon players in
the world - an artificially intelligent player called TD-Gammon. At its easiest
skill levels, OS/2 TD-Gammon will help you learn the ancient and intriguing
game of backgammon as if you were playing against a novice at the game. At its
highest skill levels, only a few of the best backgammon players in the world
stand a chance of beating TD-Gammon consistently.
Are you up to the challenge?
Further information is available on:
o What is TD-Gammon?
o Starting the OS/2 TD-Gammon program
o How to play OS/2 TD-Gammon
o Setting the skill level
o Starting a new game
o Exiting the program
o Rules of backgammon
ΓòÉΓòÉΓòÉ 1.1. What is TD-Gammon? ΓòÉΓòÉΓòÉ
TD-Gammon is a world-famous computer program that has revolutionized the game
of backgammon. Based on breakthrough research in Artificial Neural Network
technology, TD-Gammon actually taught itself to play backgammon, starting from
scratch. It learned to play so well that it is now one of the very best
backgammon players in the world.
OS/2 TD-Gammon is a special edition of TD-Gammon that uses the facilities of
OS/2 to bring this world-class game to you. It was developed by Gerald Tesauro
and Keith Weiner at the world-famous IBM Thomas J. Watson Research Center,
where it started as an experiment in artificial intelligence.
Tesauro started with an artificial neural network, a computer simulation
inspired by the human brain, that knew only the basic rules of backgammon: what
moves were legal, when the game is over, how to keep score, and so on.
He then put this neural network through a "training procedure" in which it
played many games against itself, and learned from the outcome of each game.
At the start of training, the network only knew how to choose randomly from the
list of legal moves. Over the course of many thousands of training games, the
network gradually improved its playing ability. After millions of games of
experience, the neural network is now playing at the level of today's top human
experts.
Now you can play against this amazing computer opponent. Good luck!
Further information is available on:
o For further reading
ΓòÉΓòÉΓòÉ 1.1.1. For further reading ΓòÉΓòÉΓòÉ
There are many good books available on backgammon strategy. Two-time World
Backgammon Champion Bill Robertie wrote a particularly fine book for beginners:
Bill Robertie, "Backgammon for Winners," Cardoza Publishing, ISBN:
0-940685-42-6.
Bill Robertie has also written an article describing his experience in playing
against TD-Gammon:
B. Robertie, "Carbon versus Silicon: matching wits with TD-Gammon," Inside
Backgammon, Volume 2, No. 2, pp. 14-22 (1992). Inside Backgammon is a
newsletter which can be obtained from The Gammon Press, P. O. Box 294,
Arlington, MA 02174.
A detailed technical article by Dr. Gerald Tesauro, the IBM Research scientist
who developed the neural network used by TD-Gammon, describes the techniques it
uses:
G. Tesauro, "Temporal Difference Learning and TD-Gammon," Communications of the
ACM, Volume 38, No. 3, pp. 58-68 (March 1995).
ΓòÉΓòÉΓòÉ 1.2. Starting the OS/2 TD-Gammon program ΓòÉΓòÉΓòÉ
To start OS/2 TD-Gammon, double-click on its icon. Or, from an OS/2 session,
switch to the directory in which OS/2 TD-Gammon is installed and type TDGAMMON.
For instance, if OS/2 TD-Gammon is installed in C:\GAMES, you would type:
[C:\] CD \GAMES
[C:\GAMES] TDGAMMON
A window will appear containing a graphical display of a backgammon board set
up in the standard starting configuration. TD-Gammon has the blue checkers,
and they move in the counter-clockwise direction. You have the red checkers,
and they move in the clockwise direction. Your home board is the lower left
quadrant, and TD-Gammon's home board is the upper left quadrant.
ΓòÉΓòÉΓòÉ 1.3. How to play OS/2 TD-Gammon ΓòÉΓòÉΓòÉ
You play OS/2 TD-Gammon just like a regular backgammon game, except that you
use the mouse to move your pieces on the board. You move your pieces, and OS/2
TD-Gammon moves its pieces.
Further information is available on:
o The starting board position
o Rolling your dice
o Moving your checkers
o Offering a double
o Accepting or declining a double
o Keeping score
o Rules of backgammon
ΓòÉΓòÉΓòÉ 1.3.1. The starting board position ΓòÉΓòÉΓòÉ
At the start of each game, the doubling cube has a value of 1, and is
positioned in the middle, indicating that either player can double. Also at the
start of each game, the players each roll a single die to determine who goes
first. The player who rolled the higher value gets to play the combination.
Your dice are red, and TD-Gammon's dice are blue. If TD-Gammon gets to play
first, you will see it make a move, and then the dice will disappear. If you
get to play first, you must make a legal move corresponding to the dice
combination that is shown.
ΓòÉΓòÉΓòÉ 1.3.2. Moving your checkers ΓòÉΓòÉΓòÉ
Checkers are moved by clicking on them with the left mouse button, dragging
them to the desired location, and then releasing the mouse button. If you
attempt to move to an illegal location, the computer will beep and the checker
will reappear at the starting location.
If the computer has hit any of your checkers, they will be on the bar (the gray
strip in the middle of the board). Checkers on the bar must re-enter the board
in the opponent's home quadrant before other checkers can be moved.
Once you have gotten all your checkers into your home quadrant, you can begin
bearing them off the board. To bear checkers off, drag them to the gray border
area on the left edge of the board. Checkers that have been borne off are
shown stacked up edgewise.
Before you have finished making a complete legal move, you are allowed to take
back some or all of the component moves that you have made so far. To take
back a move, simply drag the checker back to the starting location.
ΓòÉΓòÉΓòÉ 1.3.3. Offering a double ΓòÉΓòÉΓòÉ
Each time that it is your turn to play, you can offer a double if the doubling
cube is in the middle or on your side of the board. To offer a double, click on
the doubling cube with the left mouse button. If TD-Gammon accepts the double,
the cube will appear on TD-Gammon's side of the board, and will be set to twice
the previous value. If TD-Gammon declines the double, a message will appear
indicating that you have won the game; congratulations! Once you have doubled,
you cannot double again, as TD-Gammon has the exclusive right to make the next
double.
If you do not wish to offer a double, simply click anywhere on the board, and
your dice roll will appear. Once your dice have been rolled, you cannot double
until your next turn.
ΓòÉΓòÉΓòÉ 1.3.4. Accepting or declining a double ΓòÉΓòÉΓòÉ
When TD-Gammon offers you a double, the cube will be turned to twice its
previous value, and will be moved toward your side of the board. To accept the
double, click on the doubling cube and drag it to the notched cube-holding
location on your side of the board. To decline the double, click on the
doubling cube and drag it towards the middle of the board.
ΓòÉΓòÉΓòÉ 1.3.5. Keeping score ΓòÉΓòÉΓòÉ
OS/2 TD-Gammon keeps a running total of your score in the lower right-hand
corner of the window. You start out with a score of 0. When you win a game, the
number of points you win in that game is added to your score. If TD-Gammon wins
the game, the number of points it wins is subtracted from your score.
If you have won more points that TD-Gammon, your score will be positive
(greater than zero). If TD-Gammon has won more points, your score will be
negative (less than zero).