Greg Strong wrote on Fri, May 20, 2005 03:11 PM UTC:
Mark:
Your idea is very clever, and deserves an 'A' for inginuity! This will
allow a person to play synchronous chess against Zillions so long as
computer is White. But there's a problem... When the computer considers
what move to play, ('thinking',) it is recursively looking at hundreds
of thousands of sequences of move-counter-move combinations to determine
which is best. During this look-ahead, whether Zillions is White or
Black, Zillions is playing both sides and uses perfect information. The
way Zillions decides on moves will not change from a non-synchronous game
(and I'm pretty confidant that sometimes different moves are better in
synchronous chess; and if not, then what's the point?) So, essentially
what you have is a way for a person to play synchronous chess with the
computer, but the computer is still just playing chess.
Derek:
It is almost certainly possible to write a program to do it... (not that I
know how to go about it...) But the suggestion you make of a computer vs.
computer synchronous match has an additional nasty complication that is
really hard to explain, but I will give it a shot.
To have computer vs computer synchronous, you need not one capable
program, but two seperate (and different) programs. Here's why: say you
try to do it with one program... You give it the ability to handle the
hidden information by not including any code that looks at variables that
it's not supposed to. Ok, so far, so good... So, now it must try to
'guess' what the other player is going to do. Chess programs all do
this by assuming that the opponent will make the best move he can. In
this case, the 'best move he can' determination is being made by the
program! After thinking about it, the program is going to determine that
the best move is always the actual move! So, you've slowed it all way
down by making it think about the same things over and over again, but you
haven't changed its play at all! It's still just playing regular Chess
against itself... Wierd, huh?!?
Mark: Your idea is very clever, and deserves an 'A' for inginuity! This will allow a person to play synchronous chess against Zillions so long as computer is White. But there's a problem... When the computer considers what move to play, ('thinking',) it is recursively looking at hundreds of thousands of sequences of move-counter-move combinations to determine which is best. During this look-ahead, whether Zillions is White or Black, Zillions is playing both sides and uses perfect information. The way Zillions decides on moves will not change from a non-synchronous game (and I'm pretty confidant that sometimes different moves are better in synchronous chess; and if not, then what's the point?) So, essentially what you have is a way for a person to play synchronous chess with the computer, but the computer is still just playing chess. Derek: It is almost certainly possible to write a program to do it... (not that I know how to go about it...) But the suggestion you make of a computer vs. computer synchronous match has an additional nasty complication that is really hard to explain, but I will give it a shot. To have computer vs computer synchronous, you need not one capable program, but two seperate (and different) programs. Here's why: say you try to do it with one program... You give it the ability to handle the hidden information by not including any code that looks at variables that it's not supposed to. Ok, so far, so good... So, now it must try to 'guess' what the other player is going to do. Chess programs all do this by assuming that the opponent will make the best move he can. In this case, the 'best move he can' determination is being made by the program! After thinking about it, the program is going to determine that the best move is always the actual move! So, you've slowed it all way down by making it think about the same things over and over again, but you haven't changed its play at all! It's still just playing regular Chess against itself... Wierd, huh?!?