24 Jul 06
I don't know, maybe some day engines will realize that 1. e4 move for white is a winner anyway you play it, so we will have to put restrictions to the depth of the machine, and maybe no human will be able to win against a machine that has all the posibilities recorded on a hard drive.
I understand that posibilities in a chess board are a lot, but finite, so in some time engines will have them all.
the first pc I owned had 1G hard drive, and I though that I would never be able to use so much space, and now........
24 Jul 06
Originally posted by SupermanFinite maybe but it might as well be infinite.
I don't know, maybe some day engines will realize that 1. e4 move for white is a winner anyway you play it, so we will have to put restrictions to the depth of the machine, and maybe no human will be able to win against a machine that has all the posibilities recorded on a hard drive.
I understand that posibilities in a chess board are a lot, but finite, ...[text shortened]... ad 1G hard drive, and I though that I would never be able to use so much space, and now........
There are 20 possible 1st moves for white and 20 possible replies - 400 possibilities.
By the end of move 3 with a typical opening (Ruy Lopez) the number of possibilities has exceeded 236million.
Very rapidly we approach numbers that exceed the number of atoms in the known universe. Way beyond the memory capabilities of any known projections of computing power.
24 Jul 06
Originally posted by Dragon Firebut there are much fewer than 10^73 possible positions on the board. (64!/32! for arranging 32 pieces on 64 squares, 2 for whose turn and 5^16 for various possiblities for each pawn [that is PQRNB] and finally 2^30 each piece is either on or of the board)
Finite maybe but it might as well be infinite.
There are 20 possible 1st moves for white and 20 possible replies - 400 possibilities.
By the end of move 3 with a typical opening (Ruy Lopez) the number of possibilities has exceeded 236million.
Very rapidly we approach numbers that exceed the number of atoms in the known universe. Way beyond the memory capabilities of any known projections of computing power.
but this includes positions that are illegal and clearly won positions not to mention a large number of duplicate positions (swapping two white pawns creates the same position as does swapping two pieces and then removing them from the board.) I would approximate somewhere between 10^20 and 10^30 unclear positions that need analysis and an equal amount that are obviously won/drawn/lost for white. If you want to create a perfect chess machine you would use a positional tablebase not try to brute force it from move 1.
24 Jul 06
Originally posted by aginisnot to mention the many times you can move Nf3 then Ng1
but there are much fewer than 10^73 possible positions on the board. (64!/32! for arranging 32 pieces on 64 squares, 2 for whose turn and 5^16 for various possiblities for each pawn [that is PQRNB] and finally 2^30 each piece is either on or of the board)
but this includes positions that are illegal and clearly won positions not to mention a large number of ...[text shortened]... erfect chess machine you would use a positional tablebase not try to brute force it from move 1.
25 Jul 06
Originally posted by FabianFnasThat is why they were talking about modifing this rule in engine chess, so it could resign because of mate in 133 or so, maybe a human can't do it, but a machine clearly does.
I still say it is impossible due to the 50-move rule.
Mate earlier than 50 moves or else it is a plain draw.
25 Jul 06
Originally posted by SupermanThere isn't a federation of engine chess so I don't know exactly who is doing this talking.
That is why they were talking about modifing this rule in engine chess, so it could resign because of mate in 133 or so, maybe a human can't do it, but a machine clearly does.
25 Jul 06
Originally posted by SupermanMy first computer had a 20 MB harddrive and 1 MB of RAM. My current box can see a little more than twice the ply-depth. When we measure RAM in GB and harddrive space in terabytes, we'll again have doubled the depth of view, which should add another 500 to the ELO of the leading engines.
I don't know, maybe some day engines will realize that 1. e4 move for white is a winner anyway you play it, so we will have to put restrictions to the depth of the machine, and maybe no human will be able to win against a machine that has all the posibilities recorded on a hard drive.
I understand that posibilities in a chess board are a lot, but finite, ...[text shortened]... ad 1G hard drive, and I though that I would never be able to use so much space, and now........
Engines are a long way from "solving" chess. Play on.