Deep_Junior (C)
(2702)
- GM G. Kasparov
(2845)
[B44]
X3D "Man vs. Machine" Match
New York, NY (Round # 4), 02.02.2003
[A.J.G.]
Yet another key game.
I did not get to watch this game
live, (as it was being played, my
computer had
crashed); but it
seemed to me that Garry was
reeling after his loss. (In game #3.)
He plays the opening well, and there
seems to be many chances for at
least a try
for an advantage for Black. (For instance, Black by-passes several opportunities
for the sharp ...b5.)
In the end, he dilly-dallies too long, and Deep Junior will not destroy its
position
as Deep Blue did. The
program strikes out with a Q-side
pawn advance, and
only a SUPER-HUMAN defense by Black allows
him to draw this game.
In the end, the programmers had to
be happy with this effort by their
creation, but
Garry could only be
shaking his head and wondering
where he went wrong.
All
the diagrams here are inverted - just for a change.
(I wanted you too see the board from Kasparov's
viewpoint.)
1.e4, {Diagram?}
Deep Junior - good or bad - always
plays 1.e4 in this match.
[
In the Kramnik vs. Deep Fritz
match, the move 1.d4!?,
{Diag?}
saw
a lot of use. ]
1...c5;
{Diagram?}
In all three games as Black, Garry
has elected to use the Sicilian
Defense.
(This was written after
the match was over.)
The Sicilian Defense ... is very old,
being played in game over
300
years ago, and maybe a whole
lot older.
(It was mentioned in the
famous "Polerio Manuscript" that
was
first 'published' in 1594.)
Greco analyzed this opening, and
it was even played by many others.
(Notably in the famous series of
matches between La Bourdonnais
and
MacDonnell in 1834.)
But the
real father of the Sicilian would
have to be one Louis Paulsen.
***
Louis Paulsen, (1833-1891); was
one of the first great players of the
near modern
era. A contemporary
of Morphy ... he was ranked among
the world's best 6 or so
players, from
the late 1850's until nearly the 1880's.
He was a true pioneer in the
opening,
enriching dozens of lines with his
ideas. He also played nearly all the
older forms of the Sicilian.
(His use
of these lines were characterized
by an
early
...e6; and then even
a later ...a6.
He also would mix
in ...Qc7; and ...Nc6.)
He was also
notoriously slow, taking sometimes
hours on a single move.
(His rather
extensive 'thinking time' is credited by
some for 'forcing' the use
of chess
clocks in chess tournaments.)
Paulsen was also a great defensive
player,
even Steinitz would exhaust
himself in futile efforts to defeat him.
Paulsen won or
placed near the top
in over a dozen international
tournaments.
(His greatest success
may have been Leipzig 1877. He
won clear first and carried
off a
brilliancy prize for his game vs.
Zukertort.) He also won many
matches.
Towards the end of his
career, he played less.
[
Interesting would have been the
try 1...e5!?;
{Diagram?}
to see what lines
the players would have chosen. ]
2.Nf3, {Diagram?}
White controls the center and
this also develops a piece.
(Additionally, it prepares K-side
castling and helps to cover many
key squares.)
This is the move that is most often
played and leads to the main lines;
at least in this variation.
[
White could play: 2.Nc3,
{Diagram?}
which could lead to the lines
of the Closed Sicilian. ]
2...Nc6;
{Diagram?}
This is the older method of playing
the Sicilian.
(The modern method
is 2...d6.)
Of course there is absolutely nothing
wrong with this move. It is just that
(I think)
Garry is trying to avoid reams
of analysis that the Deep Junior team
may have
had months to prepare.
[
The move 2...d6!?;
{Diagram?} is much more fashionable by
current GM standards.
(It also may be a tad more flexible.)
{See Game # 6.}
]
Both sides - for the next couple
of moves - use the most often
played line.
(d4, usually on the
third move, leads to what is
known as an "Open Sicilian.")
3.d4 cxd4; 4.Nxd4 e6; {Diagram?}
Now by transposition, we have
reached what is today known as:
"The Taimanov Variation."
{See MCO-14; pages 300-316.}
(---> The Taimanov is basically
a sub-variant of the Paulsen.)
GM Taimanov was a gifted player,
{also a concert pianist}; who was
also very
creative. He was a fairly
strong player who won the USSR
Championship in 1956.
He was
also easily in the 'Top Ten' in the
world for most of decade of the
1950's.
n the late 60's and early
70's, he began playing chess
with renewed vigor, winning
3
straight international tournaments.
He then qualified for the Candidate
Matches,
but had the misfortune
of being paired against Fischer -
who was making his
successful
run at the World's Championship.
Poor Taimanov was flattened 0-6.
(This effectively ended his career.)
According to MCO, this is not the
'correct' move order for this line,
but I don't think it really matters.
(MCO gives: 1.e4, c5; 2.Nf3, e6;
3.d4, cxd4; 4.Nxd4, Nc6.)
So it is probably a near-meaningless
transposition here.
(I am sure that Kasparov had a
reason for choosing this move
particular order,
but I have little
concrete ideas as to exactly why.)
5.Nb5!?,
{Diagram?}
This is not the main line.
(Nc3 is the move that is most
often played in Master praxis.)
Of course there is nothing wrong
the move, Nb5 here. It is aggressive
and designed
to force Black to play
an early ...d6; thus making Black
close off his dark-squared
Bishop
from any attempts at early play.
*******
[
MCO gives the variation of:
5.Nc3 a6; 6.g3,
{Diagram}
This is the main line here;
and a favorite in Master play
since Fischer used it to defeat
GM M. Tal
in Bled, 1961.
(Also playable is: 6.Be2!?, "+/=")
6...Qc7;
{Diagram?}
This is the main line, at least
according to one book I have -
and also Modern Chess Openings.
( Black could also play: 6...d6; 7.Bg2 Bd7!?; 8.0-0 Nf6;
9.Nxc6! Bxc6;
10.a4! Be7; 11.a5 0-0; 12.Be3 Nd7; 13.Na4,
"~" {Diagram?}
This is a position that slightly
favors White, he has a development
and a spatial advantage here.
K. Guseinov - P. Tregubov;
/Dub. Open Tournament/ Dubai UAE 2002 (1-0, 50) )
(We now return to the MCO line.)
7.Bg2 Nf6; 8.0-0 Be7;
9.Re1 Nxd4; 10.Qxd4 Bc5; 11.Qd1,
11...d6;
12.Na4 Ba7; 13.c4! Bd7; {Diagram?}
The end of the column.
14.Be3 Bxe3; 15.Rxe3 0-0;
16.Nc3 Qxc4; 17.Qxd6, "+/=" {D?}
White has a very small edge
here.
Djuric - Gostisa;
Yugoslavia, 1991.
[ See MCO-14; pages # 302,
column #1, and notes # (a.)
through (f.). ]
***
White could try: 5.c4!?,
{Diagram} as well, in this position.
(Several of my older books highly
recommend this line. But this
analysis
is very old, and based on
books from the late sixties and also
the very
early seventies.)
]
*******
The next few moves appear
to be relatively forced. (best)
5...d6; 6.c4 Nf6; 7.N1c3 a6; 8.Na3 Nd7!?;
('!') {Diagram?}
A strategic withdrawal, but other
than that, I am not sure what the
purpose of this
move is. It also
has the benefit of getting out of
book as quickly as possible.
I am sure that many wanted to see
Garry play the adventurous ...d5;
which led
to a complete rout of
Anatoly Karpov in the 16th game
of their second World
Championship
Match. (1985) But there is no need
for such a wild play ... anyway,
the
DJ team would obviously be prepared
for this.
(Theory has passed this line by,
it is little played today.)
*******
[
MCO gives the continuation of: 8...d5!?;
9.cxd5 exd5; 10.exd5,
10...Nb4;
11.Be2 Nfxd5; {Diagram}
This is supposedly forced.
(11...Bc5!?; 12.Be3!, "+/=")
12.0-0 Be6; 13.Qa4+! b5;
14.Naxb5 axb5; 15.Bxb5+ Ke7;
16.Nxd5+ Nxd5;
17.Qe4, "~" {Diagram?}
And White has great compensation for the material.
GM Z. Almasi - Horvath;
National Championship,
Budapest, HUN; 1993.
[ See MCO-14; page # 307, column # 16, and note # (o.). ]
***
The main line is: 8...Be7;
9.Be2 0-0; 10.0-0 b6!?; 11.Be3 Bb7;
12.Qb3 Nd7;
13.Rfd1 Nc5; {Diagram}
The end of the column.
14.Qc2 Qc7; 15.Rac1 Rac8;
16.Nab1 Nb4; 17.Qd2 Qb8;
18.f3
Rfd8; "=" {Diagram?}
MCO calls this equal, I am more
inclined to award a VERY small
advantage to White. ("+/=")
GM L. Yudasin - GM J. Lautier;
Pamplona, (ESP); 1992.
[ See MCO-14; page # 307; column # 16, and note # (p.). ]
]
*******
For the next nine (or so) moves,
both sides develop in a relatively
normal fashion.
9.Nc2 Be7;
10.Be2 b6!?; 11.0-0 Bb7; 12.h3!?, {Diagram?}
What does this move accomplish?
(One commentator on one server
gave this move a whole question
mark.
As it did not significantly
worsen the program's position, I
think that is both wild and {too} extreme.)
[
Probably better is: >/=
12.b3
]
12...0-0; 13.Be3 Rc8;
14.Qd2 Nce5!?; 15.b3 Nf6;
16.f3 Qc7; 17.Rac1!? Rfe8!?; {See
the diagram just below.}
Evaluation time. According to
many different programs, White
has an
extremely small advantage
in this position. But with exact play,
I feel
that Garry should be OK.
My only question is:
Why didn't Garry play the freeing
pawn advance of ...b5?
(One server makes the comment
here that Kasparov is content to wait.)
I think the move ...Rfd8;
would also have been much more natural than
moving the rook to e8.
[
I think a big improvement (over
the game) would be the line:
>/=
17...b5!; 18.f4!?,
{Diagram?}
This could be forced.
( Not 18.cxb5?? Qxc3; "/+" Or 18.Rb1!?
bxc4; "=/+" )
18...Ned7; 19.f5 Ne5;
20.fxe6 fxe6; 21.Nd4 b4; "=/+"
(Maybe - "/+")
{Diagram?} (Analysis
line.)
Black is CLEARLY for choice
in this position.
(I spent several
DAYS looking at these lines and
searching for moves,
all with
the help of several different
computer programs.) ]
18.a3!?, {Diagram?}
What the heck is DJ doing here?
(The computer seems adrift here.)
[
Maybe better is: 18.a4,
"=" {Diagram?} to discourage ...b5. ]
18...Ned7!?;
{Diagram?}
This is OK, but ...
I am sure against virtually any
human opponent, Garry would
hardly
hesitate
now before playing
the move ...b5!
[
Why not: 18...b5!;
"=/+" ? {Diagram?}
(Black is at least a little better.)
]
19.Rfd1 Qb8!?; {Diagram?}
Here Garry is playing a strategy
like 'rope-a-dope' ... something
that worked
very well against the
Deep Blue program.
(But here it comes very close to
back-firing completely!)
[
Black could consider: 19...Nc5;
20.Rb1 Red8; "<=>" {Diagram?}
with fair play.
]
20.Bf2 Rcd8;
('?!') 21.b4!? Ba8!?; {Diagram?}
Garry - VERY uncharacteristically -
begins to play very passively.
[
Maybe a slight improvement
over the game would be:
>/=
21...Rc8; 22.Na4 Bc6; 23.b5 axb5;
24.cxb5 Ba8;
25.Nb4 Nc5!;
26.Na6 Qa7; "~" {Diagram?}
when Black has achieved a
fortress-like position that
he should not lose.
]
22.a4! Rc8;
23.Rb1 Qc7; {Diagram?}
Garry has adopted a very passive
type of 'do-nothing' strategy. And
after missing
several opportunities
for a freeing break on the Q-side,
he can do little but wait
for the
program to find the best line.
24.a5!,
(Maybe - '!!')
{See the diagram just below.}
An excellent move, especially
for a computer. The machine
is willing to
(temporarily) sack
at least one pawn for a complete
Queen-side bind.
(A couple of GM's thought this
move was bad, and {incorrectly}
predicted the
machine would lose
horribly with this move.)
Several GM's who were watching
this game were already predicting
this move
at this point in the game.
(World #3 Vishy Anand was following
along at Playchess.com
and he
immediately
said that he had been
wondering if this ambitious push was
possible.)
[ 24.Rb2!?
]
24...bxa5; 25.b5,
{Diagram?}
The only correct follow-up to
the program's entire idea.
[ 25.bxa5?
Nc5; "=/+" {Diag?}
]
25...Bb7;
26.b6 Qb8; {Diagram?}
Some had predicted that White's
b-pawn would be terribly weak
here and prone to capture ... but this turned out to be nothing
more than (really) wishful thinking!
*****
Over the course of the next
10 - to - 15 moves, dozens of
annotators and GM's
professed
that they had no real idea what
was happening in this game!!
(This included GM's Seirawan
and Maurice Ashley, the official
commentators
for this match.)
27.Ne3 Nc5!?; {Diagram?}
Pronounced as essential by some ...
and as useless by others!!
[
Maybe possible was: 27...d5!?;
"~" {Diagram?}
with maybe a playable game
for Black. ]
28.Qa2 Nfd7;
29.Na4 Ne5!?; "~" {Diagram?}
I must say - that having spent WEEKS
annotating this game,
(I saved this
one for last!) - that it is VERY hard to determine who
is better here!
[
Several programs recommend
that Black play: 29...h5!?;
{Diag?}
(Maybe - '?!') but this looks too weakening to me. ]
Both sides continue to maneuver.
From the confused and sometimes
conflicting
comments - many of
the GM's were unsure who was
better here.
30.Nc2 Ncd7; 31.Nd4 Red8; 32.Kh1 Nc6; 33.Nxc6 Rxc6;
34.Kg1 h6;
35.Qa3 Rdc8; 36.Bg3 Bf8; 37.Qc3 Ne5!?; {Diagram?}
This strikes me as rather too
committal. Maybe Black should
delay any
clarification, his position
rather resembles a fortress.
(I believe Garry was very short
of time here.)
***
[
Maybe a more prudent course
for Black would have been:
37...Be7;
38.Qa3, {Diagram?}
It looks better to delay the
capture of the a-pawn until
a later time.
(Or 38.Qxa5!? Bf6; "~")
38...f6; 39.Bf2 Ne5; "~"
{Diagram?}
when Black's play against
White's c-pawn off-sets his
other problems in this position.
]
38.c5!?, {Diagram?}
The analysis from an unidentified
author on the ChessBase web site,
{http://www.chessbase.com/games/2003/x3d4.htm}; calls this dubious.
('?!')
Another author brands this as
"terrible" and awards this a whole
question mark. ('?')
Personally, I feel both of these are
questionable judgment calls and
are very
reactionary. Weeks of
analysis has failed to find a forced
win - at least one
that is relatively
easy to find or 100% forced.
Therefore - while I am slightly
inclined to believe this might be a
premature attempt at a break-through, I
cannot prove anything.
And in the complete absence of a
verifiable forced
win for White, a
great deal of consideration must
be given to the player that
DJ is
facing in this game. (Kasparov!)
[
ChessMaster 6000
likes: 38.h4!?,
('?!') {Diagram?}
but after ...Nxc4; I think that
Black is at least slightly better. ("=/+")
***
Maybe best for White would have
been the continuation of:
>/=
38.Bxe5 dxe5; 39.Qxa5,
"+/=" {Diagram?}
when White is clearly at least
a little better.
(But this is not
an iron-clad conclusion. Black
could gain a
lot of play later,
on White's greatly weakened
dark-squares.) ]
38...Nd7;
39.Qxa5 Nxc5; 40.Nxc5 Rxc5; 41.Qa4 R5c6!; {Diag?}
Garry has decided to return
material to reach a theoretically
drawn game ...
something the
box is not yet smart enough to
figure out on its own.
(At least not in advance.)
42.Bf2 d5;
(!) 43.Bxa6 Bc5;
44.Bxc5 Rxc5; 45.Bxb7!?, {Diagram?}
Was this best? It is the choice
of several strong programs. But
Garry seems
to think that it was
better to retain the light-squared
Bishop on the board.
[
Was 45.exd5!?, {Diagram?}
better? ]
45...Qxb7; 46.exd5 exd5;
{Diagram?}
This is probably forced.
[
Black could lose quickly with: 46...Ra8?;
47.Qxa8+! Qxa8;
48.b7 Qb8;
49.d6! Rd5!?; 50.Rdc1!, ("+/-")
{Diagram?}
and White has a won game. ]
47.Qa7 R5c7!; {Diagram?}
The best move. Its also a cute trick.
(The computer would never fall for
it, but a human in time
pressure
might.)
[
Black could also play: 47...Rb8!?;
48.Qxb7 Rxb7;
49.Kf2 Rc6!;
{Diagram?} which seems to be basically
the same idea as the game. ]
48.Qxb7, {Diagram?}
White wins a pawn. The poor
dumb box thinks it is winning.
[
Not 48.bxc7?? Qxa7+; ("-/+") {Diagram?}
]
The rest of the game, while
interesting to some is deadly
dull. The computer
still might
be rendering evaluations of
"White is winning."
(Or at least "±" which means that White
is clearly better.) But the real
truth
is that the computer could
maneuver until h____ freezes
over ... and not
make any real
progress here.
***
Kasparov soon has his watch back
on ... and is giving other very clear
indications that this game is over.
(The program is the last to
know in this game!)
48...Rxb7; 49.Rxd5 Rc6; 50.Rdb5 h5; (!) 51.Kf2 Re6;
52.f4 g6; 53.Kg3,
53...Kg7; 54.Kh4 Kh6; 55.R1b4 Rd6; 56.g3 f6;
(!) 57.g4,
{Diagram?}
Does the computer think it
is actually making progress
in this position?
[
Kasparov could only pray for
a line like: 57.Rb3!? Rc6;
58.R5b4?? g5+;
59.fxg5+ fxg5#
]
57...hxg4; 58.hxg4 Kg7;
59.Rb3 Rc6; 60.g5 f5;
61.Rb1, {Diagram?}
DRAW
White cannot take his Rooks off
the b-file .......... and therefore
is
unable to make any progress.
Here the programmer actually figured
out that the program was NOT
able to
really improve its position ...... and
either offered a draw here,
or accepted Kasparov's offer of a draw
at this point.
In the end, it may have been this
titanic struggle, (close to 6 hours);
that sapped so much of Garry's
energy. (In the next game, Garry
allows the computer to escape after
playing basically an unsound
sacrifice. This is EXTREMELY rare
for a player of Kasparov's caliber!)
A day or two after this game, I
was saying (on ICC) it looked like
Garry was
probably lost in this game.
An IM called me down for this remark,
saying that
Kasparov was never in
real danger of losing this game. To
disprove this, I
offer the following
quote:
"Kasparov admitted after the
game that had Junior kept the
light-squared
Bishops" {on the board} "his position
was probably lost."
(From the CB
web site.)
The
final position in this contest.
A strange game, the computer
alternated really stupid moves
with really great
ones on an almost
regular basis.
Copyright (c) A.J. Goldsby
I. Copyright (c) A.J.G; 2003.
(Code initially) Generated
with ChessBase
8.0
½ - ½
|
