White: Deep Blue
|
Black: Garry Kasparov
|
1.e4
Deep Blue sticks to the opening move which proved so successful in
game 2. |
|
|
c6
Is Kasparov going to play his first love the Caro-Kann Defence? |
2.d4
A logical space-gaining move. |
|
|
d6
What is this? The commentators were amazed see 2...d6. |
3.Nf3
IM Summerscale, one of Britain's leading exponents of this defence
says- 'This is not the strongest move'. This defence which is known as
the Czech or Pribyl system gives White the chance to build a broad pawn
centre with f4, so preferable was 3.Nc3 keeping the option of f4 open.
John Henderson commenting on the choice of opening said: "The Prybl
system is not-such a bad idea against a computer. As white, Kasparov can
set the agenda and choose what he wants to play. We have to remember that
Deeper Blue has a vast opening book, and Kasparov's job is to find a way
around it."
"His 3rd game was a classical example of this. By playing 1 d3 and 4
a3 it "confused" the computer, who misplaced the bishop on e7 allowing
Garry to transpose into a favourable English Opening."
"The interesting point is if Deeper Blue has such a sophisticated opening
book then why didn't it notice the possibilty of a transposition into an
English?"
"However, as Black, Garry's job is much, much more difficult. The "tin
box" will automatically play 1 e4. Garry doesn't want to get involved in
a tactical battle with a Sicilian. Remember, in the last match that was
his downfall in the first game; he tried to out-calculate a super-computer
in a tactical position, which is akin to challenging a Ferrari in a roadrace
with a Sinclair C5!" |
|
|
Nf6
A developing move attacking White's e4-pawn. |
4.Nc3
Developing and defending at the same time. |
|
|
Bg4
A clever move planning to erect a pawn barrier on the white squares
e6, d5 and c6. But first bringing the white-squared bishop outside the
pawn chain. |
5.h3
The normal continuation, putting the question to the bishop. Do you
exchange the bishop for the f3-knight or retreat the bishop to h5? |
|
|
Bh5
Keeping the tension in the position. |
6.Bd3
An alternative is 6.Qe2, which is the most aggressive move in this
position - preparing to castle queenside and attack on the kingside. GM
Hebden from England popularised this system. |
|
|
e6
This is the next stage in Kasparov's plan, putting his pawns on light
squares now that his bishop is outside the pawn chain. |
7.Qe2
White keeps the option of castling queenside, while playing a move
that strengthens his e4-square. |
|
|
d5
Kasparov stakes his claim in the centre, as outlined in previous moves. |
8.Bg5
This appears to be a new move. The only move we can find in our database
is 8.exd5. White counters the growing pressure on his centre by pinning
the black knight on f6. (8.exd5 8...Bxf3 9.Qxf3 cxd5 10.0-0 (10.Ne2 Nc6
11.c3 a6 12.0-0 Rc8 13.Ng3 Be7 14.Bg5 0-0 15.Rae1 Ne8 16.Bc1 g6 17.Bh6
Ng7 18.Qe3 Re8 19.Re2 Bh4 20.Qf4 Na5 21.Qg4 Papaioannou,I-Oratovsky,M/EUchJ
Vejen 1993/½-½ (64)) 10...Nc6 11.Ne2 Be7 12.c3 0-0 13.Bf4 a6 14.Ng3 b5
15.a3 Na5 16.Qe2 Bd6 17.Bxd6 Qxd6 18.f4 Nc4 19.Rf3 Rfe8 20.Raf1 Rac8 21.R1f2
1-0 Chernyshov,K-Maiwald,J/Dresden op 1995 (39)) |
|
|
Be7
Breaking the pin.
|
9.e5
A commital move, but one which side-steps any problems with Black playing
Nxe4. Not 9.0-0? which is a good example of the type of tricks waiting
for Deep Blue: 9...Nxe4!! 10.Bxe7 Nxc3 11.Bxd8 Nxe2+ 12.Bxe2 Bxf3 13.Bxf3
Kxd8 Leaving Black with a winning position. |
|
|
Nfd7
Black is preparing to break back in the centre with c5. |
10.Bxe7
Removing what is commonly known as Black's good bishop which is not
obstructed by any pawns. However exchanges generally help Black in this
position since he has less space for his pieces than White, therefore it
is possible that the plan of 8.Bg5 is inaccurate. |
|
|
Qxe7 |
11.g4
An aggressive move from the computer, planning to castle queenside. |
|
|
Bg6
|
12.Bxg6
A double-edged move. 'This is dubious, better would have been 0-0-0,
followed by h4 and h5 forcing ...Bxd3 when the recapture cxd3 strengthens
White's centre.' - GM McDonald. |
|
|
hxg6
Black now has a far more compact pawn formation on the kingside since
White's pawns on g4 and h3 look a bit loose. |
13.h4
White starts a far-sighted plan to open up the kingside. |
|
|
Na6
Preparing to castle queenside. |
14.0-0-0
Not only castling the king into safety, but connecting the rooks. |
|
|
0-0-0 |
15.Rdg1
'I prefer Black in this position' - IM Summerscale. 'I think it is
roughly level' - IM Tozer. |
|
|
Nc7
A intelligent re-routing of the knight. However Kasparov spent approx.
35 minutes on this move! |
16.Kb1
A standard defensive move, removing the White king from the c1-h6 diagonal. |
|
|
f6
Beginning the attack on White's centre. Time left Kasparov 52mins,
DeepBlue 1 hour 34 mins. |
17.exf6 |
|
|
Qxf6
'This is very surprising, I expected 17...gxf6.' - GM McDonald. Although
if here 17...gxf6 then 18.g5 Rdf8 and Black is doing well.
Joe Hoane from the DB Team commenting said: "The computer will never
play perfect chess. That's not the goal of this project, because it's not
possible to play perfect chess. In the last six months, the computer has
gone from playing [human-type] chess every once in a while to playing chess
almost all the time." |
18.Rg3
White anticipates pressure along the f-file and so bolsters his knight
on f3. |
|
|
Rde8
Preparing to blast open the centre with 19...e5. |
19.Re1
Deep Blue has 1:24 to Garry's 0:49.
'Black still has a clear plus' - IM Summerscale. Laying siege to the
e5-square and so preventing the e5-break. Deep Blue wants to gain solid
contol of the position and then try to exploit some structural weaknesses
in Black's kingside. However White also has potential weaknesses in his
position e.g. h4, f4 and the f-file in general. |
|
|
Rhf8
Another option was 19...Re7 with the idea of doubling on the e-file
and trying to play e5. |
20.Nd1
Hiarcs thinks White stands better. |
|
|
e5!?
'Bizarre, Kasparov has gone beserk' - GM McDonald. 'An interesting
positional pawn sacrifice' - IM Summerscale. 'Why didn't Kasparov play
20...Qf4 first' - IM Tozer. |
21.dxe5
The computer is only too willing to accept material. |
|
|
Qf4
Black has sacrificed a pawn, hoping to recover it in the next couple
of moves, when his superior co-ordination of pieces will give him the advantage.
But it is not so clear how he can in fact regain the material since the
e5-pawn is heavily fortified. Kasparov is clearly speculating. The whole
match may be decided over the next couple of moves. |
22.a3
Kasparov is shaking his head. The World Champion cnnot believe this
move. Has the computer got no respect? Instead 22.h5 looks much better,
when if 22...g5? 23.Qe3 attacks a7 and the g5 pawn is doomed. |
|
|
Ne6
This is the perfect blockading square for the knight, however it cuts
off the black rook's attack on the e5-pawn. |
23.Nc3
Re-centralising the knight. |
|
|
Ndc5
Black's pieces are finding good squares, but does he have enough compensation? |
24.b4
'A Terrible positional move, ask the c4-square what it thinks about
the move b4.' - IM Summerscale. 'A human would never play that.' - IM Tozer. |
|
|
Nd7
Kasparov retreats his knight and is delighted to hav provoked the rash
b4 move. Perhaps Deep Blue should play 25 Qe3 offering the exchange of
queens which it would be difficult for Kasparov to refuse in view of the
attack on a7. Then the weakness of his king's position would cease to be
of such significance. |
25.Qd3
Attacking g6. |
|
|
Qf7
Defending f6. |
26.b5
Trying to justify b4, but it cedes the c5-square to a black |
|
|
Ndc5
Kasparov jumps in at once, taking advantage of White's last move. |
27.Qe3
Also possible was 27.Qd2. |
|
|
Qf4
Garry offers to trade queens since he is possibly afraid to cross swords
with the computer in a sharp middlegame especially with time-trouble looming.
Once the queens are exchanged White will have strong pressure down the
f-file to compensate for the pawn deficit. |
28.bxc6
It is not possible for Black to exchange queens yet due to the intermezzo
29.cxb7 check! |
|
|
bxc6 |
29.Rd1
Preventing any tricks with d4. |
|
|
Kc7
Kasparov now has 19 minutes for 11 moves, while the computer has 51
minutes. Kasparov anticipates an endgame and so improves the position of
his king. This quiet move also clears the b-file for the rook and gives
Deep Blue the chance to make the positional blunder 30.Qxf4 Rxf4 when the
f2-pawn is very weak. Kasparov now has 19 minutes for 11 moves, while the
computer has 51. |
30.Ka1
Removing the White king away from any danger on the open b-file. |
|
|
Qxe3
'I'm very suspicious of the White e-pawns' - IM Summerscale. If 30...Qc4
then 31.Nd2 Qa6 32.Ne2 with a slight advantage for White. |
31.fxe3 |
|
|
Rf7
Preparing to double on the f-file. |
32.Rh3
'I can't see the point to that move, surely there is something more
constructive. I think White should have played h5 at some point to break
things open on the kingside.' - GM McDonald. |
|
|
Ref8
Intending to invade along the f-file. White's position is beginning
to creak. |
33.Nd4
White's best chance of safety is to challenge the knight on e6. |
|
|
Rf2
The invasion begins. |
34.Rb1
Can the computer generate enough counterplay on the queenside. |
|
|
Rg2
Threatening to double rooks on the seventh rank which in combination
with Nxd4 would win the c-pawn, when White's position would crumble. |
35.Nce2
Deep Blue prefers to sacrifice the g-pawn, rather than the disastrous
position given in the note above. |
|
|
Rxg4
Kasparov regains the pawn and leaves White's pawn structure shattered. |
36.Nxe6 |
|
|
Nxe6 |
37.Nd4
This is the best fighting chance to gain some counterplay in a rook
ending. As we know 'All rook endings are drawn'. |
|
|
Nxd4
We feel that Kasparov has at least a draw in the coming rook endgame
with good winning chances. The key to his position is gaining passed connected
c- and d-pawns. If he is able to achieve this he has tremendous winning
chances. |
38.exd4 |
|
|
Rxd4
Picking up the loose d-pawn. |
39.Rg1
Deep Blue wants to capture both the g-pawns and place a rook on the
seventh rank. |
|
|
Rc4
Garry is optimising the activity of his pieces, while trying to gain
the key connected c- and d-passed pawns. |
40.Rxg6 |
|
|
Rxc2 |
41.Rxg7+
The pawns are coming off, but the passed c- and d-pawns are stronger
than the three isolated white pawns. |
|
|
Kb6 |
42.Rb3+ |
|
|
Kc5!
A much better move than 42...Ka6, which would hold the a7-pawn. The
key to the position is keeping the key active. |
43.Rxa7 |
|
|
Rf1+
A strong move forcing White to place his rook passively on the back
rank. |
44.Rb1
The Deep Blue team offered Garry a draw at this moment, Garry just
laughed and said 'let's play chess'. |
|
|
Rff2
Garry can now force a draw at any moment with perpetual check from the
rooks along the second rank. But Garry undoubtedly wants more. |
45.Rb4 |
|
|
Rc1+ |
46.Rb1 |
|
|
Rcc2
If instead 46...Rxb1+ then 47.Kxb1. |
47.Rb4
Is Garry content with a draw? |
|
|
Rc1+
Alternatively 47...Ra2+ 48.Kb1 Rxa3 49.Rxa3 Kxb4 is another way to
try and play for a win. |
48.Rb1+
No! Garry will play for the win. |
|
|
Rxb1+ |
49.Kxb1 |
|
|
Re2 |
50.Re7 |
|
|
Rh2 |
51.Rh7 |
|
|
Kc4
Not as good is 51...d4. It is better to keep the king in front of the
pawns. This will enable Black to threaten back rank checkmate as well as
trying to promote his pawns. |
52.Rc7 |
|
|
c5 |
53.e6 |
|
|
Rxh4 |
54.e7 |
|
|
Re4 |
55.a4 |
|
|
Kb3 |
56.Kc1
Draw Agreed |
|
|
|
½ |
½ |