I've heard it said by more than one commentator that computers are too
heavily biased by material to sacrifice it for positional advantages that
have no immediate tactical payoff. It puzzles me when they say that because
I've seen CM9K do it quite often. Here's an example (also attached). The
program sacs one pawn and later offers more. The later ones may be tactical,
but the pawn I took on my (black) move 14 doesn't fit what these
commentators say. I've come to the conclusion that Chessmaster 9000
evaluates positions and attack potential very well beyond the material
advantage that is farther down the road than the machine is looking.

[Event ""]
[Site ""]
[Date "2003.8.24"]
[Round ""]
[White "Chessmaster"]
[Black "Wilma"]
[TimeControl "40/7200"]
[Result "1-0"]

1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.O-O Nxe4 5.Re1 Nd6 6.Nxe5 Be7 7.Bd3 Nxe5
8.Rxe5 O-O 9.Nc3 c6 10.Re1 f5 11.Qf3 Bg5 12.b3
Qf6 13.Rb1 Qh6 14.Bf1 Bxd2 15.Bxd2 Qxd2 16.Re3 Qd4 17.Rd1
Qf6 18.Bd3 g6 19.Bf1 Nf7 20.Ne4 Qb2 21.Nd6 Nxd6 22.Rxd6 Qg7 23.a4 Kh8
24.Qe2 a6 25.Re7 Rf7 26.Rxf7 Qxf7 27.Qe5+ Qg7 28.Rf6 1-0

"Wilma" wrote in message
nk.net...
I've heard it said by more than one commentator that computers are too
heavily biased by material to sacrifice it for positional advantages that
have no immediate tactical payoff. It puzzles me when they say that
because
I've seen CM9K do it quite often. Here's an example (also attached). The
program sacs one pawn and later offers more. The later ones may be
tactical,
but the pawn I took on my (black) move 14 doesn't fit what these
commentators say. I've come to the conclusion that Chessmaster 9000
evaluates positions and attack potential very well beyond the material
advantage that is farther down the road than the machine is looking.

[Event ""]
[Site ""]
[Date "2003.8.24"]
[Round ""]
[White "Chessmaster"]
[Black "Wilma"]
[TimeControl "40/7200"]
[Result "1-0"]

1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.O-O Nxe4 5.Re1 Nd6 6.Nxe5 Be7 7.Bd3 Nxe5
8.Rxe5 O-O 9.Nc3 c6 10.Re1 f5 11.Qf3 Bg5 12.b3
Qf6 13.Rb1 Qh6 14.Bf1 Bxd2 15.Bxd2 Qxd2 16.Re3 Qd4 17.Rd1
Qf6 18.Bd3 g6 19.Bf1 Nf7 20.Ne4 Qb2 21.Nd6 Nxd6 22.Rxd6 Qg7 23.a4 Kh8
24.Qe2 a6 25.Re7 Rf7 26.Rxf7 Qxf7 27.Qe5+ Qg7 28.Rf6 1-0


I was taken by the statement "evaluates positions....farther down the road
than it is looking" Which, as you probably understand, is impossible.

So I decided to investigate this game, especially the move of interest, to
see if the sacrifice was a) a sacrifice b) sound. I have come up with some
quick observations.

Regarding the first question, is it a sacrifice...typically I have worked
with the strict definition of sacrifice being the loss off a piece for
positional gain. A pawn is not a 'Piece' under that definition, and while
I'm not a GM, like Kasparov, I place a value of less than 100 points, in
general, for a pawn. So no, I think a reasonable person can conclude it's
not a bona fide sacrifice. But of course, others might (And do) say any
purposeful loss of material is a sacrifice.

Then comes the better question, is it a sound move/plan for white. Well, I'm
not exactly sure. Granted a few things happened that are beneficial for
white, the open lines, the loss of tempi for black, the trading of the weak
first rank bishop for the well placed black bishop, and a possible winning
attack on the F pawn, if so desired. All in all, not unreasonable for the
loss of 100 points.

Maybe more information is required as to how CM9000 was configured for this
game, was it a personality that valued pawns less, like Kasparov? Was it
forced to move quickly? Are you running it on an underpowered PC, where ply
is limited by the time alotted?

I've run this position on a few PC's different chess engines, all robust,
and white does have a slight positional advantage, by the numbers, by the
moves made, to around 14 play, however, it's very slight. So there's no
'magic' invovled, it's number cruching with non-intuitive results.

I did find this interesting, and I might look at it more in depth, I wish I
had a copy of CM9K...I bet I can find it locally.

Interesting. As to the setup, it was the Chessmaster personality itself
without modifications. The time control was 40 moves per 2 hours. The
machine is a Pentium 4, 1.6 gB. However, I made the first 8 moves for both
sides before giving CM the white pieces.

Wilma

"Derek Wildstar" wrote in message
news:f3R2b.202351$cF.66559@rwcrnsc53...

"Wilma" wrote in message
nk.net...
I've heard it said by more than one commentator that computers are too
heavily biased by material to sacrifice it for positional advantages
that
have no immediate tactical payoff. It puzzles me when they say that
because
I've seen CM9K do it quite often. Here's an example (also attached). The
program sacs one pawn and later offers more. The later ones may be
tactical,
but the pawn I took on my (black) move 14 doesn't fit what these
commentators say. I've come to the conclusion that Chessmaster 9000
evaluates positions and attack potential very well beyond the material
advantage that is farther down the road than the machine is looking.

[Event ""]
[Site ""]
[Date "2003.8.24"]
[Round ""]
[White "Chessmaster"]
[Black "Wilma"]
[TimeControl "40/7200"]
[Result "1-0"]

1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.O-O Nxe4 5.Re1 Nd6 6.Nxe5 Be7 7.Bd3 Nxe5
8.Rxe5 O-O 9.Nc3 c6 10.Re1 f5 11.Qf3 Bg5 12.b3
Qf6 13.Rb1 Qh6 14.Bf1 Bxd2 15.Bxd2 Qxd2 16.Re3 Qd4 17.Rd1
Qf6 18.Bd3 g6 19.Bf1 Nf7 20.Ne4 Qb2 21.Nd6 Nxd6 22.Rxd6 Qg7 23.a4 Kh8
24.Qe2 a6 25.Re7 Rf7 26.Rxf7 Qxf7 27.Qe5+ Qg7 28.Rf6 1-0


I was taken by the statement "evaluates positions....farther down the road
than it is looking" Which, as you probably understand, is impossible.

So I decided to investigate this game, especially the move of interest, to
see if the sacrifice was a) a sacrifice b) sound. I have come up with some
quick observations.

Regarding the first question, is it a sacrifice...typically I have worked
with the strict definition of sacrifice being the loss off a piece for
positional gain. A pawn is not a 'Piece' under that definition, and while
I'm not a GM, like Kasparov, I place a value of less than 100 points, in
general, for a pawn. So no, I think a reasonable person can conclude it's
not a bona fide sacrifice. But of course, others might (And do) say any
purposeful loss of material is a sacrifice.

Then comes the better question, is it a sound move/plan for white. Well,
I'm
not exactly sure. Granted a few things happened that are beneficial for
white, the open lines, the loss of tempi for black, the trading of the
weak
first rank bishop for the well placed black bishop, and a possible winning
attack on the F pawn, if so desired. All in all, not unreasonable for the
loss of 100 points.

Maybe more information is required as to how CM9000 was configured for
this
game, was it a personality that valued pawns less, like Kasparov? Was it
forced to move quickly? Are you running it on an underpowered PC, where
ply
is limited by the time alotted?

I've run this position on a few PC's different chess engines, all robust,
and white does have a slight positional advantage, by the numbers, by the
moves made, to around 14 play, however, it's very slight. So there's no
'magic' invovled, it's number cruching with non-intuitive results.

I did find this interesting, and I might look at it more in depth, I wish
I
had a copy of CM9K...I bet I can find it locally.