Divockreni
03-21-2004, 06:53 PM
I just thought of something and I wanna know if it's at all possible. Chessmaster.com has an EGDB for download that can make Chessmaster play perfectly when six pieces remain on the board (as if it's not perfect already http://ubbxforums.ubi.com/infopop/emoticons/icon_smile.gif. The screen might say, "Mate in 89" or something ridiculously cool like that. I generated a 5-piece database, and it took a few days straight to build. The 6-piece takes at least 1024 gygs to build and 1 YEAR on a 1.5 ghz comp (according to the readme).
If a six-piece database can be created, how about a 32-piece? It would have to calculate ALL the possible combinations of moves in the game of chess! Maybe I should e-mail someone from Chessmaster or the EGDB creator. The EGDB works by a scoring system and "distance to mate". The question is, would it be able to score moves from the beginning of the game, when each player is in equal position?
It sounds possible to me. I think it could score the moves the same way as before but based on a LOT more tests. There would be no distance to mate in the first move (unless chess is flawed...not possible!), but black would have to respond with certain moves else be checkmated (my guess is that for both white and black there would be more than one perfect move at times). God knows how long it would take and how much space it would require. Maybe if we teamed supercomputers (can we even do that?) to build it we could get it done in say, 10 years? Or do I underestimate chess and/or overestimate supercomputers? What would be hilarious is if a grandmaster were to play it and early in the game see, "Mate in 277".
Of course, this might not be necessary because the developers of Deep Blue might have found an easier means to perfection. Not that Deep Blue is perfect, but perhaps the process by which it calculates its moves can be improved a little bit until there is a perfect computer opponent. The problem is we wouldn't KNOW that it's 100% perfect, whereas with an *Entire* GDB we would because it uses "brute force" calculation--it's got every game in chess memorized.
Peace without wisdom is as fragile as comfort. Wisdom without peace is not wisdom.
If a six-piece database can be created, how about a 32-piece? It would have to calculate ALL the possible combinations of moves in the game of chess! Maybe I should e-mail someone from Chessmaster or the EGDB creator. The EGDB works by a scoring system and "distance to mate". The question is, would it be able to score moves from the beginning of the game, when each player is in equal position?
It sounds possible to me. I think it could score the moves the same way as before but based on a LOT more tests. There would be no distance to mate in the first move (unless chess is flawed...not possible!), but black would have to respond with certain moves else be checkmated (my guess is that for both white and black there would be more than one perfect move at times). God knows how long it would take and how much space it would require. Maybe if we teamed supercomputers (can we even do that?) to build it we could get it done in say, 10 years? Or do I underestimate chess and/or overestimate supercomputers? What would be hilarious is if a grandmaster were to play it and early in the game see, "Mate in 277".
Of course, this might not be necessary because the developers of Deep Blue might have found an easier means to perfection. Not that Deep Blue is perfect, but perhaps the process by which it calculates its moves can be improved a little bit until there is a perfect computer opponent. The problem is we wouldn't KNOW that it's 100% perfect, whereas with an *Entire* GDB we would because it uses "brute force" calculation--it's got every game in chess memorized.
Peace without wisdom is as fragile as comfort. Wisdom without peace is not wisdom.