Question by Living L: How a computer chess program works ?
I have always wondered what kind of algorithms are employed on the computer chess program that can easily win the best players in the world.
Does it uses a database which have set of all possible combinations and the best step stored in it ? Nope it must not be like that because that program size will be extremely large in that case.
Or are there many rules in the game of chess that if followed we can always win. For example in Tic Tac Toe game, learning some tricks always makes you atleast tie with the opponent or even win if your opponents dont know those.
Chess is one of the most logical game in the world. What is the secret beyond it AI ? Across speed and accuracy what makes a chess program the best chess player ?
Best answer:
Answer by Tizio 008
chess is not like tic tac toe: special winning sequences does not exists (at the beginning and in the most common situation you can’t force your enemy to do exactly something … when you can do it, the game is at an end… and likely your enemy will resign before the chessmate can become reality, since at that point he/she can see he/she can’t survive)
there were several efforts to create a “thinking machine” that tried to work like a human thinking on a chessboard. all failed.
as far as i know, the winning approach is a database of known winning and loosing positions (taken from played chess tournaments), a set of rule to respect (the chess rules…) and a very powerful supercomputer that tries a lot of combinations of moves in a depth search. smarter algorithms “cut” the dead leaves of the depth search tree, so they can analyse more moves spending less time in loosing or poor moves.
but basically all is still based on a depth search where a lot of moves are “Imagined”, and the one with the best continuation is taken (the best continuation is the one that makes it win, or take a good position, according to the known positions in the database or an evaluation of the chessboard … an evaluating algo can be added as an “help” for the depth search algo, and it is necessary to “cut” bad leaves or branches …)
more sophisticated algo are adaptive, they can learn step by step… if you do a good move and you win, next time you won’t 
IBM DeepBlue
http://en.wikipedia.org/wiki/Deep_Blue
won Kasparov, but was hard…
when i play against gnu chess, i loose 99/100 … i think deepblue can kill gnu chess on an intel p4 machine using 1/100 (or less) of its power…
Add your own answer in the comments!
Incoming search terms:
- the secret of learning chess
- computer chess
- chess learning from database
- step by step chess moves to win computer
- step by step chess learning pc
- learning from the database in computer chess
- kill computer chess
- is computer chess killing chess?
- how a chess program works
- computer chess learning with database
