Kasparov-Deeper Blue, Match 1997
Deeper Blue Swindles Draw in Game Five
Kasparov opened the contest in the same vein as the first and third games:
attempting to exploit Deeper Blue's overt preference of bishops to knights.
I believe everyone was surprised by 4. .. Bxf3; in all probability,
the machine realised that the bishop could come off if White desperately
wanted it after 4. .. Bh5 6. g5 Bg6 7. Nh4. Of course, in this line
White's position is quite weak, being far too extended on the kingside
and completely devoid of development elsewhere on the board.
7. .. Ne5 is an interesting bit of play, commencing a sequence wherein
White's bishop gets kicked around quite a lot in the early going.
Deeper Blue's play is quite ironic, as it in fact demonstrates the
kind of opening strategy and aesthetic supposedly reserved for Kasparov.
However, the first real shot was fired with 11. .. h5!?,
an unorthodox move to say the least.
Kasparov wears his emotions on his sleeves, and his look of
surprise matched the buzzing amazement of the audience.
At first blush it seems to be a farcical move, but the machine's
tactical resourcefulness correctly indicates that White's
kingside should be pressured.
Eventually Kasparov was somewhat forced to justify the
machine's impertinence with the blocking move 15. h4.
This allowed Deeper Blue to open the game with 15. .. e5,
and suddenly the character of the game seemed to favor the machine.
After 22. .. N6e5, Black lays the trap
23. Rxh5? c5! 24. Nf3 Rxd1+ 25. Qxd1 Nxf2.
But 25. .. Nc4?! fails to exploit Black's potential to play
c5-c4 and gain an outpost for his knight on d3.
At this point Kasparov had twenty-nine minutes for his next
fifteen moves; Deeper Blue had an hour and eleven minutes,
though Mike Valvo had earlier stated that twenty minutes was alloted
for disaster time.
Nonetheless, tournament players know this is crunch time;
Kasparov made the forty-move time control in less than fifteen
26. Qa4 is a world-class crunch-time move,
especially with against an opponent you are sure can out-calcuate you
(as Kasparov admitted in the post-game conference).
Kasparov deftly mixes tactical threats in order to coordinate
his pieces before Black could further rend his position.
27. Re1 was necessary as 27. Rxd8 Qxd8 28. Qxc4?? Qd1 mates.
But the pawn at f2 is still weak.
Deeper Blue then discards all advantage with 29. .. Qg6?,
when 29. .. Qd3 could hold a small advantage, even after the
White rook invades at e7. Though Black is up a pawn after
30. .. Nxf2, its kingside pawns are christmas candy whereas
White's are not so vulnerable.
35. Bd5 and 36. Re6 prevent the knight at f2 from escaping to e4,
but the astute shot 36. .. Nb5 not only eliminates bishop for
knight (if 37. Kb2 Nd3+), but weakens the White king's defense.
With 39. .. Ne4, Black is clearly looking for a draw by perpetual
check. Perhaps 47. Re2 was worth a shot.
But in the end it seemed there was no way to stop Black from
shuttling his rook between d2 and d1, giving check all the while.
White: Garry Kasparov
Black: Deeper Blue
Game Five of Six, 10 May 1997
1.Nf3 d5 2.g3 Bg4 3.Bg2 Nd7 4.h3 Bxf3 5.Bxf3 c6
6.d3 e6 7.e4 Ne5 8.Bg2 dxe4 9.Bxe4 Nf6 10.Bg2 Bb4+
11.Nd2 h5 12.Qe2 Qc7 13.c3 Be7 14.d4 Ng6 15.h4 e5
16.Nf3 exd4 17.Nxd4 O-O-O 18.Bg5 Ng4 19.O-O-O Rhe8 20.Qc2 Kb8
21.Kb1 Bxg5 22.hxg5 N6e5 23.Rhe1 c5 24.Nf3 Rxd1+ 25.Rxd1 Nc4
26.Qa4 Rd8 27.Re1 Nb6 28.Qc2 Qd6 29.c4 Qg6 30.Qxg6 fxg6
31.b3 Nxf2 32.Re6 Kc7 33.Rxg6 Rd7 34.Nh4 Nc8 35.Bd5 Nd6
36.Re6 Nb5 37.cxb5 Rxd5 38.Rg6 Rd7 39.Nf5 Ne4 40.Nxg7 Rd1+
41.Kc2 Rd2+ 42.Kc1 Rxa2 43.Nxh5 Nd2 44.Nf4 Nxb3+ 45.Kb1 Rd2
46.Re6 c4 47.Re3 Kb6 48.g6 Kxb5 49.g7 Kb4 Draw Agreed
41.Bxg4 Bxg4 42.Nxg4 Nxg4 43.Rxg4 Rd5 44.f6 Rd1 45.g7 1-0
Annotations by Inside Chess' GM Yasser Seirawan
Nonetheless, I was distressed by the analysis of game five's final position.
Even IBM's transcript on the Web reveals audience members calling
out moves for further analysis, to no avail.
Why would no commentator analyze the move that obviously defeats the perpetual check??
Final position, Game Five; White to play
After 50. Re2:
- 50. .. Kc3 51. Rxd2 Kxd2 52. g8Q Kd1 53. Kb2
- 50. .. Rd8 52. Nd5+
- 50. .. Rd1+ 51. Kb2 (not 51. Kc2?? c3+ and mate to follow)
51. .. Rc1 52. Rc2 c3+ 53. Rxc3 Rxc3 54. Nd5+, 55. Nxc3, and 56. g8Q
51. .. Na1 (intending 52. .. c3 mate) 52. Nd5+ (intending 53. Nc3) Rxd5
53. g8Q c3+ 54. Ka2 Rd2+ 55. Rxd2 cd 56. Qg4
51. .. Nc1 52. Kc2
51. .. Ra1 52. Nd5+ intending 53. Nc3
51. .. c3+ 52. Kc2
52. .. Nd4+ 53. Kxd1 Nxe2 54. Nxe2
52. .. Rd2+ 53. Rxd2
White wins in all the above lines.
The best try for Black is 50. .. Rd1+ 51. Kb2 c3+ 52. Kc2 Rc1+
53. Kd3 Rd1+ 54. Ke3 Rd8.
[In this line, if 51. .. Rc1, 52. Nd5+ and 53. Nc3 secures the win.]
I strongly felt that White should win this position,
but analysis shows that Black's threats are sufficient to draw:
White to play; Game Five, extrapolated position
- 55. Nd3+ Kc4
(55. .. Rxd3+ looks promising but fails after 56. Kxd3 Nc1+
57. Ke3 Nxe2 58. Kxe2 Kb3 59. g8Q+ or 58. .. c2 59. Kd2)
56. Ne5+ Kb4 57. Rf2 Rg8 58. Rf7 c2 59. Rxb7+ Ka5
- 55. Nd5+ Kc4 56. Ne7 Nd4 57. g8Q Rxg8 58. Ng8 c2
- 55. Ne6 Rg8
56. Rf2 Kc4 57. Rf8? c2
56. Rc2 b6 57. Kd3 Nc5+ 58. Nxc5 bxc5
- 55. Rf2 Re8+ (demanding that the White king return to his box)
56. Kd3 Nc5+ 57. Kc2 Rg8 58. Nd5+ Kc4 59. Ne3+ Kb5 60. Rf7 Ne6
61. Nf5 Kc4 62. Rxb7 a5, deflecting the White rook from the defense
of g7, draws.
Eventually White cedes the g7 pawn
when the Black king cedes the c3 pawn to assail the g7 pawn,
and the extra pawn at g3 is not sufficient to win.
The Black king guards g8, and its rook harasses White's king
from the far rank and the queen's wing.
I had planned to be in New York City that day anyway, so I decided
to see if I could attend my first big chess event. After walking from
the Upper West Side and down 7th Avenue during pleasant weather,
I turned up at around half-past two in the afternoon with no ticket but cash.
The one thing about New York, no matter who you are, money moves the mama.
There were a few guys hanging around front, looking for tickets.
I went inside to a large, forty-foot high lobby with a
circular seating area in the middle; on the other side of the circle across
from the main entrance were two metal detectors flanking a standard 6'x3'
convention table, no ticket, no entrance.
I wandered back out front, and began the "extra ticket to sell?" mantra,
joining three others on a large and (for the Apple) lightly-travelled sidewalk.
One guy got lucky while my back was turned. After eight or ten minutes, some
dude comes out from the lobby, clearly pleased there's a market brewing.
He's slighter than me, maybe 5'8", 140 pounds, same ugly brown hair as me.
With the glasses the stereotype suggests white middleclass techie, bred and raised.
He asks my price; I venture forty (for a $25 ticket) and knock out the guy next to me
who joined me about two minutes into my mantra and resolved to pay face value.
So the hawker then says, wait a minute, and walks halfway down the block to the guy
who was there before me for his bid.
The hawker clearly scoped the scene.
The guy who was there before, an amiable dude, offers sixty.
I, trailing the hawker, immediately counter sixty-five.
The other guy goes right to eighty.
The hawker says, ok, whoever gives me a hundred first gets the ticket.
I pull a c-note out the small pocket of my levis,
simultaneously snag and swap the ticket for the bill.
The hawker and I leave together and head toward the lobby.
As I enter the revolving door, the dude apologizes for squeezing me.
Right. The punk's probably on the IBM payroll, I think.
I pass the metal detectors,
and the fiftysomething security guard wants to inspect my pack,
which for some reason doesn't go through the detectors.
He doesn't like the bag;
there are a few backpacks entering, but none stuffed like mine.
He makes me open it, like it's an airline flight (remember
when they didn't even ask for domestic flights?? or am i really old at
He notices the chess set, clock, and books,
but scowls at the smaller black bag,
demands to know "what's in there??" like immigration.
I smile, open the smaller bag while saying it's my shaving
kit--actually, all of the toiletry junk I carry now that I'm old.
Off I go, down a small flight of stairs.
I feel like I'm going to a movie, no ambience at all.
There's another, low-celing lobby, and from there you enter the
auditorium. It looks like an oversized version of an MIT lecture hall.
I grab a seat in the last row of the center section, one seat away
from the aisle.
The guy next to the aisle is amiable,
has a freshly-purchased copy of Godel's Proof,
the same edition which I read some twelve years ago.
He won't mind if I mutter to myself.
The show begins.
On stage there are three moderators:
Mike Valvo, a master who's head of the ACM's chess division;
IM Maurice Ashley, an impeccably-dressed black guy
with an excellent speaking voice;
and GM Yasser Seirawan,
well-known to me as the top U.S. player during the eighties and still
one of the better ones.
Two huge tv screens flank a projection of the
single laptop that accompanies the commentators.
The left screen shows
the room where Kasparov is playing; the right screen is a closeup
of the game board.
The laptop has an interesting piece of chess software that allows
the chess experts to demonstrate possible variations in the game.
To the audience's right on stage is a duplicate of the machine named
RS/6000 SP some 35 floors up.
It's not doing much, just sitting there,
draped in a bland powder blue
some eight feet high and four feet square.
Not even any blinking lights, just two rows of six or eight LEDs.
As the game progressed, the commentary team kept pace with the action
on the laptop, exploring variations and defending their individual
take on the game to the audience. After the first half-hour or so,
one of the team would take a break and another noted chess personality
would take his place.
Of particular intellect and entertainment
was the Women's World Champion Zsuzsu Polgar,
who handled the audience's questions superbly and
was even more impressive discussing the game's play-by-play.
A troupe of elementary school chess-players were in attendance,
and the requsite q&a fit in well without irritating pandering.
I was privileged to be recognized by GM Seirawan and
ask a question of the IBM rep who came out.
I inquired whether the team had anticipated addressing
Deeper Blue's rather static time management in the future
(it only takes two to four minutes a move,
whereas Kasparov would take ten seconds to sixteen minutes in the game,
and some GMs, particularly Korchnoi and Bronstein, have spent more than
an hour on a single move in certain games).
I mentioned the Fischer clock in passing.
I've always thought that if time management was accounted for,
then perhaps chess programs could regain the high road from
chess expert systems to artificial intelligence--at least
time management was applicable in other expert systems.
Unfortunately, IBM sent a PR guy who had some responsibility for
the staging and little else,
offered lots of excuses,
and not someone who had any sense of the underlying software.
He punted, and Mike Valvo gave an irrelevant answer in his place.
This was consistent with IBM's public face;
no truly relevant technical insight as to the workings of
Deeper Blue were ever expressed by a member of the team
or even ghost-writer.
In fairness, of the four hundred or so audience members,
my question shot over the head of all but fifty to eighty.