Originally posted by rubberjaw30Maybe computers are "so far behind in their positional understanding" because it is not necessary to have a lot of positional understanding in order to win a game of chess! Humans rely on positional understanding because they can't number-crunch efficiently enough.
none of you have answered my question:
why are computers so far behind witht their positional understanding?
for example: in the King's Gambit Accepted, don't computers give black a plus, when history shows that white has plenty of compensation for the pawn, and black usually has to give the pawn back to equalize?
The most brilliant and creative human player who devises beautiful and profound strategic conceptions will get slaughtered by the best computer program, because chess is almost 100% about number-crunching!
Originally posted by gaychessplayernot true! the best computers don't consistently beat the best players.
Maybe computers are "so far behind in their positional understanding" because it is not necessary to have a lot of positional understanding in order to win a game of chess! Humans rely on positional understanding because they can't number-crunch efficiently enough.
The most brilliant and creative human player who devises beautiful and profound stra ...[text shortened]... laughtered by the best computer program, because chess is almost 100% about number-crunching!
Originally posted by tomtom232No top player has defeated a top computer in match play in over a decade!
not true! the best computers don't consistently beat the best players.
Hydra defeated one of the top ten players in the world, GM Michael Adams, by a score of 5 wins, 1 draw and 0 losses.
The best computer absolutely, positively SLAUGHTERS the best human players.
Over ten years ago, Gary Kasparov lost a match to Deep Blue. Kramnik drew a match several years ago, and the computer he played wasn't even the world's best computer program!
Someone please tell me when was the last time a GM won a match against a top computer program.
Originally posted by gaychessplayermy bad! I guess I should get myself more up to date.
No top player has defeated a top computer in match play in over a decade!
Hydra defeated one of the top ten players in the world, GM Michael Adams, by a score of 5 wins, 1 draw and 0 losses.
The best computer absolutely, positively SLAUGHTERS the best human players.
Over ten years ago, Gary Kasparov lost a match to Deep Blue. Kramnik drew a ...[text shortened]... Someone please tell me when was the last time a GM won a match against a top computer program.
Originally posted by rubberjaw30it's because even 20 plies is still far from endgame, it's often not even enough to see the opening through. a human knows from experience that "this kind of position will be very hard to defend" etc. without knowing the specific moves. for computer such things are beyond its horizon, totally hidden. it's just poking in the dark with a 20 plies deep stick, and it's not even checking out all continuations but disregarding (pruning) some of them with a general 'rule of thumb', which the programmer has guessed to work adequately in an average situation.
none of you have answered my question:
why are computers so far behind witht their positional understanding?
for example: in the King's Gambit Accepted, don't computers give black a plus, when history shows that white has plenty of compensation for the pawn, and black usually has to give the pawn back to equalize?
in a nutshell: human understands, the computer just adds with no knowledge of what it is actually doing.
Originally posted by gaychessplayerIf "number crunching" was all that determined a good chess engine there would be little difference between software packages and we would long ago have seen engines as strong as the best ones in existence today.
Maybe computers are "so far behind in their positional understanding" because it is not necessary to have a lot of positional understanding in order to win a game of chess! Humans rely on positional understanding because they can't number-crunch efficiently enough.
The most brilliant and creative human player who devises beautiful and profound stra ...[text shortened]... laughtered by the best computer program, because chess is almost 100% about number-crunching!
While calculational power is important in a chess engine, the heuristics programmed into it are what determines is strength or weakness. In fact, the first chess engines were calculation heavy and heuristic weak, and played very poorly.
Even today's best chess software, run on fast parallel processors and with endgame tablebases and opening databases available to them, do not have the time to carry-out systematic and exhaustive calculations of all possible lines from start position to end of game. Nor are their calculations comprehensive even within the calculation window (ply depth) used. Heuristics represent the embodiment of human positional knowledge in algorithmic form; but to the extent that this embodiment -- this reduction -- is shallower and/or less flexible, its defects must be compensated (to the extent possible) by other factors (see below). Remember that calculations must be goal directed if software is to play chess well, and that the mere gain of material, in and of itself, is inadequate to generally determine the fitness of moves, though under many circumstances it is important.
So, what is reflected by the improvement of chess engine software is really four factors: (1) improved calculation speed; (2) better heuristics; (3) better opening theory and opening databases; (4) better endgame knowledge as reflected in tablebases. The latter two factors are obviously whittling away from both ends, while the first two are working primarily in the interior.
1 edit
I have a game in progress where I used this gambit variation of the Scandi (with 3...c6). I can say that I felt no advantage at all throughout the entire opening and middlegame. Of course, it doesn't help that I obviously made some inaccurate moves. I did find a decent position eventually, but that was no result of the opening.