|
GM
Garry Kasparov
(2847)
-
Deep_Junior (C)
(2599)
[D10]
X3D "Man vs. Machine" Match
New York,
NY;
USA
(Game # 3),
Jan. 30th, 2003
[A.J.G.]
Garry
dominates this contest for almost the entire length of the whole
game.
But then a slip at the end costs him the full point.
The
game starts off as a fairly standard "Queen's Gambit
Declined," and
then goes into the line none as "The Slav Defense."
1.d4
d5; 2.c4 c6; 3.Nc3 Nf6; 4.e3 e6; {Diagram?}
We
have now entered the lines
known as the Semi-Slav ...
in the Slav proper, Black will
usually play ...Bf5.
(See game #1 of this match for
a more detailed look at the Slav,
and the move order used in
this particular game.)
[
4...Bf5!?
]
5.Nf3 Nbd7;
6.Qc2, {See the diagram
just below.}
As I discussed in Game (No.) One,
this surely was part of Garry's pre-match
preparation. Not only does
he avoid the bulk of the computer's
book, he also
gets the box into a
line it really does not play too well.
The
game position - just after 6.Qc2.
This type of closed position is also
VERY suited for a long game of
positional
chess and lots of
maneuvering ... but this is not
what Garry has in mind.
[ The main line probably is:
6.Bd3 dxc4; 7.Bxc4 b5; 8.Bd3 a6;
9.e4 c5;
"<=>" {Diagram?} with good play for both sides.
{I believe this line is called:
"The Meran Variation."}
(I have several books on this
line alone.)
For more information, see the
reference book,
"Modern Chess
Openings, 14th Edition." (MCO-14)
Pages 454-470, columns # 1
through column # 30,
and the
dozens and dozens of notes.
A fairly good game, which is also
fairly recent, is the encounter:
M. Krasenkow -
D. Fridman; 4th Borowski GM, 2002.
]
6...b6!?; {Diagram?}
Either the program likes this
rather silly and superficial
move ...
or whomever was
responsible for the opening book,
chose this
move for the box.
I
am quite sure that a much better
move ... AND the main line ...
is
the move, 6...Bd6.
[ Almost certainly better is:
>/= 6...Bd6;
{Diagram?}
which has been played many
times;
(Over 2500 different times
at the master level!);
for example:
M. Vaisser - G. Flear;
National I Tournament,
Montpellier, FRA. 2003.
]
7.cxd5 exd5; 8.Bd3 Be7;
9.Bd2,
(TN?) {Diagram?}
I have a book on the Semi-Slav
that is only a couple of years old.
(It only covers 0-0 here.)
[
White could also play: 9.0-0 0-0;
10.Bd2 h6; 11.Ne2 Bb7;
12.Ng3 Re8; 13.b4,
"+/=" {Diagram?}
White has a small but steady
pull in this position.
R. Akesson - J. Eriksson;
National Championships,
Skara, Sweden. 2002.
(0-1, 39 moves) ]
9...0-0; {Diagram?}
It is taught to every student of the
game to castle as quickly as
possible
in the opening phase,
but maybe Black would have been
better off
delaying this idea here.
But on the other hand, I spent over
two hours one day analyzing the
alternatives in this position, and it
is difficult to find anything that does
not lead to a very clear advantage
for White.
[ Even worse for the second
player would have been:
</=
9...Qc7?!; 10.Rc1! Qd6; 11.e4,
"±" {Diagram?}
and White is very clearly
better ("+/") here. ]
10.g4!?,
(Maybe - '!')
{See the diagram just below.}
Garry is determined to play
this sharp and probably good
idea.
(0-0 would have
transposed into the above note.)
The
position after 10.g4.
This move leads to seemingly almost
inexhaustible complexities. And while
the programmers like to trumpet the
(seeming) fact that computers excel
at
tactics, I have no doubt that had
Garry also had access to a computer,
he
would have crushed Deep Junior
like a bug in this game.
[
10.0-0!?
]
10...Nxg4; {Diagram?}
Did Black have to capture this
pawn? The way the game went,
White gets a
nearly won game -
at least if he had chosen the
correct continuation. So maybe
this capture is dangerous.
(But then again, Black already
has definite problems
with the
defense of his h-pawn.)
[ Slightly worse for Black was:
10...Bb7?!; 11.g5 Ne8;
12.Bxh7+,
12...Kh8; 13.Bf5,
"+/=" {Diagram?}
and White keeps a very small
but clear edge in this line.
***
Maybe playable was: 10...h6!?;
11.Rg1 c5; 12.g5 hxg5;
13.Nxg5 cxd4;
14.exd4, "+/=" {Diagram?}
White is clearly for choice here,
especially seeing that the g-file
is available for operations against
the Black King. Whether or not this
is better than the game is very
hard to say.
(White will eventually
castle on the Queen-side here.) ]
11.Rg1, {Diagram?}
This is OK, but many on-line
commentators said that Bxh7+
was
much better. (I'm not sure.)
This move is definitely one of the
sharpest moves the first player
can try at this point.
[ Maybe better was:
11.Bxh7+!? Kh8; 12.Rg1 Ndf6;
13.Bd3,
"+/=" {Diagram?}
with a small but clear
advantage
for White. ]
11...Ndf6!?;
{Diagram?}
Apparently the computer thought
for several minutes before playing
this move.
This is probably wiser than maybe
weakening Black's King-side with
the pawn advance, ...h5.
[
11...h5
]
White's position now explodes with
a lot of energy - and Garry comes
very close to winning right out of
the opening phase of the game.
12.h3! Nh6; 13.e4! dxe4;
14.Bxh6!?, {Diagram?}
This looks like fun, but is it the
very best move here?
I thought while the game was being
played that Nxe4 was better for White.
Several programs suggest that the
try Bxe4 may be winning for White!
[ Probably the best line for
White here is:
>/= 14.Bxe4! Kh8;
{Diagram?} This looks forced.
(</=
14...Nxe4?; 15.Bxh6, "±")
15.Bxc6,
"+/=" {Diagram?}
with a very clear and sizeable
edge for White here.
(Maybe even "+/".)
***
Maybe also good for White was:
14.Nxe4 Kh8;
15.Bxh6 gxh6; 16.Qxc6,
"~" {Diagram?}
with MAYBE a tiny edge for
White here, but this is not
really certain. ]
14...exd3; 15.Rxg7+, {Diagram?}
Garry's fabulous instincts told
him to keep the g-file open.
[ One on-line commentator pointed
out the tricky line:
15.Bxg7!? Ng4; {Diagram?}
This is forced.
(The machine would never fall for: 15...dxc2??;
16.Bxf6+ Bg4; 17.Rxg4#)
16.Qxd3 Kxg7; 17.hxg4,
"~" {Diagram?}
with an approximately even game.
- Mig Greengard.
]
The next few moves look forced.
15...Kh8; 16.Qxd3 Rg8; 17.Rxg8+ Nxg8; 18.Bf4 f6; ('!')
{Diagram?}
...f6 is an excellent move, and keeps
the White pieces off many key squares.
(Notably g5 and e5.)
Times: White - 55:00; Black: 1:14.
**********
[
To show the reason for the near
necessity of ...f6;
I offer the following
very complicated variation:
18...c5?!;
('?') 19.Be5+ Nf6[]; {Box, Diagram?}
This is forced.
(19...f6?; ('??') 20.Ng5! Bf5; 21.Qxf5 Kg7; 22.Qxh7+ Kf8; 23.Qf7#)
20.Ng5 Qg8; 21.Nce4 Bf5;
22.Qf3 Bxe4; 23.Nxe4!!, {Diagram?}
At first this seems like a
blunder. (!)
( Also promising is:
23.Bxf6+ Bxf6; 24.Qxf6+ Qg7;
25.Nxf7+, {Diagram?}
This is probably best.
(25.Nxe4!? Re8; "<=>")
25...Kg8; 26.Qxg7+ Kxg7; 27.Nd6,
"+/=" {Diagram?}
White is clearly better.
(Maybe "+/") )
23...Qg1+; 24.Ke2 Qxa1;
25.Bxf6+ Bxf6; 26.Qxf6+ Kg8;
27.Qg5+,
("+/-") {Diagram?}
White has a winning attack, a few
minutes with any
computer program will verify this.
(This line shows the
deadly combination of
Queen-plus-a-Knight.)
( For
example: 27.Qg5+ Kf8; {Diagram?}
This could be forced.
(Much worse is: 27...Kh8?!;
28.Qe5+ Kg8; 29.Nf6+ Kg7; 30.Ng4+,
30...Kg8; 31.Nh6+ Kf8;
32.Qd6+ Kg7; {Diagram?}
This is practically forced.
(32...Ke8?; 33.Nf5, "+/-")
33.Nf5+ Kg8; 34.Qg3+ Kf8;
35.Qg7+ Ke8; 36.d5!,
"+/-" {Diag?}
White's attack crashes through ... he will pick up the Black
Rook
whenever it is convenient
for him to do so.)
28.Qh6+ Ke8; 29.Qc6+
Kf8[]; {Diagram?}
Unfortunately, this is forced.
(</=
29...Ke7??; 30.Qd6+ Ke8; 31.Nf6#)
30.Qxa8+ Kg7; 31.dxc5 Qxb2+;
32.Kf3 Qxa2;
33.c6, ("+/-")
{Diagram?}
White's extra piece and his
passed c-pawn should
decide
the game in his favor. )
*****
Black could also try: 18...Qd7!?,
{Diagram?}
with an interesting position.
]
**********
19.0-0-0,
"+/=" {Diagram?}
Garry looks to have a DEFINITE
advantage, but ... as GM Loek van
Wely
pointed out, these are the type
of positions that computers excel at.
(Wely has trained intensively with
computers the last few years.)
What Wely probably means is that
this type of position puts a VERY
large
emphasis on TACTICS ... and this is something the computers
are good
at ... and probably better
than most humans in their handling
of these types
of positions.
Mig Greengard also mentioned this
on-line as well.
(That this type of
position - with the primary emphasis
on tactics - may look
scary for DJ,
but computers usually defend this
type of position very well.)
[ 19.Qe4!?
]
19...Bd6; 20.Qe3 Bxf4;
{Diagram?}
Black has neutralized most of
White's attack in this position.
Times: White - 41:13; Black - 1:05.
21.Qxf4 Bxh3!?;
(Maybe - '!')
{See the diagram just below.}
(The
position just after Black plays 21...Bxh3.)
This capture looks very scary
and dangerous for Black ...
White gets a lot of open lines
for the attack.
(But of course
DJ feels no anxiety at all.)
I know MANY GM's who would be
VERY reluctant to grab
this pawn
against Garry Kasparov!!!
[
White gains a clear and consistent initiative in the line with:
21...b5!?;
22.d5 cxd5; 23.Rxd5 Qe7; 24.Rxb5,
"+/=" {Diag?}
with a small edge for the first
player from this position. ]
22.Rg1 Qb8!; {Diagram?}
This is the most accurate, Black
runs into problems if he allows
White's
Queen to remain on the
very strong f4-square.
Times: White - 34:00; Black: 53:46.
[
After the continuation of: 22...Qe7;
23.Qh2 Be6; 24.Nh4 Bf7;
25.Qg2 Qf8;
26.Qxc6, "+/=" {Diagram?}
the first player winds up with
a very powerful and passed
d-pawn. (Black gets his pawn
back with ...Bxa2.) ]
23.Qe3 Qd6; 24.Nh4, {Diagram?}
This is different, but I had
expected Ne4.
(Time: White - 23:16)
[
24.Ne4!?;
or 24.Qe4!?
]
24...Be6; {Diagram?}
This simple retreat is probably
the second player's best option
at this point.
Times: White - 23:16; Black - 38:00.
[
It looked like Black could (maybe)
force a draw with:
24...Qh2;
25.Nf3 Qd6; 26.Nh4 Qh2; 27.Nf3,
{Diag?}
with a draw here by the
repetition of the position. ]
25.Rh1 Rd8;
26.Ng6+ Kg7; {Diagram?}
This is Black's only legal move.
Times: W - 20:50; B - 37:00.
27.Nf4 Bf5; {Diagram?}
This is both good and forced.
[
Black should definitely NOT
play:
27...Qxd4??;
28.Nxe6+, ("+/-")
{Diagram?}
and White wins Black's Q.
]
28.Nce2 Ne7; {Diagram?}
The programmer/operator
seemed visibly relieved when
the machine
finally re-activated
this wayward steed.
[
28...a5!?
]
29.Ng3, {Diagram?}
At this point White had 16 minutes
left on his clock ...
to make it all
the way to move 40.
[
29.a3
]
29...Kh8;
{Diagram?}
This is probably the safest
move for Black.
[ Probably dubious for Black is:
</= 29...Kg8?!;
30.Nxf5 Nxf5;
31.Qe4 Qd7;
32.Nh5 Kf7; 33.Ng3 Ng7; 34.Rxh7,
"+/=" {Diag?}
and White has a strong pull
and a clear edge.
(Black
should NOT fall for ...Qxd4??;
Rxg7+, Kxg7?!; Nf5+,
which
will win easily for White.)
(34.Rh4!?) ]
30.Nxf5 Nxf5; 31.Qe4 Qd7;
{See the diagram just below.}
Black
seems to be just barely hanging on.
(The
position just after Black's thirty-first move.)
32.Rh5,
(Probably - '?')
{Diagram?}
Could this be an error?
(It would appear that it is.)
[
Apparently White can still force
a neat draw with:
32.Ng6+!
Kg7[]; {Diagram?}
This is forced.
(Not 32...Kg8?; 33.Qxf5! Qxf5; 34.Ne7+ Kf7; 35.Nxf5,
"+/-")
33.Nf4! Kh8; {Box?
Diagram?}
This looks forced.
34.Ng6+ Kg7; 35.Nf4,
"=" {Diagram?}
with an obvious repetition.
]
Now Kasparov is rather short
of time ... and the denouement
requires
only a little accurate
calculation of the tactics.
32...Nxd4!; 33.Ng6+!? Kg8;
34.Ne7+ Kf8!; 35.Nd5 Qg7; (!)
36.Qxd4 Rxd5!;
{See the diagram just below.}
White RESIGNS.
The
final position of this interesting contest.
A great day for the program ...
but a truly bad game by Garry.
White had many opportunities to
win and/or gain a serious advantage,
and he missed most of them.
This is a line that the programmers
will surely examine in great detail ...
to see if they cannot improve upon
Black's play in this line.
(Black's opening was horrific.)
Copyright (c) A.J. Goldsby I. Copyright (c) A.J.G; 2003.
(Code initially) generated
with ChessBase
8.0
0 - 1
|
