Deep Blue (2450) - Garry Kasparov (2730) 
[B22]
Philadelphia ACM match (Game # 1), 1996.

[A.J.G.]

The first game of this historic match. 

Man vs. Machine ... an absolute monster of enormous silicon power, 
vs. the reigning (chess) World Champion.  Who would prevail? 

***

My comments here are based on the book of the match;
by GM Raymond Keene, Byron Jacobs, and also T. Buzan. 

(I have many books on the first match, the one by Keene, Pandolfini, 
 Shamkovich, etc. Additionally I have been adding notes by Michael 
 Khodarkovsky, from his book: "A New Era.") 

1.e4 c5; {Diagram?} 
A Sicilian, Kasparov has no fear. 

2.c3!?, {Diagram?}   
"A sensible choice by the computer programmers." - Keene. 

He goes on to remark that Kasparov knows his main lines inside and out, 
and this will avoid much of his pre- match preparation and knowledge. 

Apparently this was before the DB team had assembled a massive opening 
book. (Like the one that the Deep Fritz team used against GM V. Kramnik.) 

     [ I would prefer the wide-open (regular) Sicilian lines: 
       2.Nf3 d6;  3.d4 cxd4;  4.Nxd4 Nf6;  5.Nc3, "+/=" {Diagram?} etc. 
       White has great play, and many opportunities to find great tactical shots. ] 

2...d5!;  {Diagram?}  
The best move - many of my older books give this an exclam. 
(It is based on the fact that White cannot play Nc3.) 

(Keene remarks that this is already a mistake. This is ludicrous. 
 Even if the game opens up, the best move is still the best move.) 

     [ The other main choice here is to play the Knight to f6. 
       The main line runs as follows:  2...Nf6!?;  3.e5 Nd5;  4.d4 cxd4;  
        5.Nf3 Nc6;  6.Bc4 Nb6;  7.Bb3, "~"  {Diagram?}   
       Theory has tended to favor White here.   (Maybe - "+/=")   
       See the contest:   M. Savic - P. Popovic;   57th National Championship. 
       Banja Kovilijaca, YUG; 2002.  (White won a model game.)   

       MCO quotes the game:  Rytshagov - M. Sadler;   Pula, 1997. ]  

3.exd5 Qxd5;  4.d4 Nf6!;  5.Nf3 Bg4!?;  {Diagram?} 
Something very new and relatively untried. 
(But Kasparov has used it at least once before.) 

I am sure Garry played this ... in the hopes that the programmers of 
his silicon opponent had not put this move into the opening book. 

This is probably a solid line, but as I often play this line as Black; 
I prefer to keep my QB at home. 

Dozens of master level games since this contest was played, proves 
that this IS a fully playable line.   

But Keene and Jacobs say this is very risky, and that the very solid 
move,  5...e6; should have been played. And while I do not agree that 
the move  ...Bg4;  is even risky - I do think ...e6!;  is a much better and  
much more solid choice for Black.   

   [ MUCH better, and much more solid than the game, is the continuation: 
      >=  5...e6!;  6.Bd3 Be7;  7.0-0 0-0;  8.Be3 cxd4;  9.cxd4 Nc6;  
      10.Nc3 Qd6;  11.a3! b6!?;  {Diagram?}  The end of the column.  
      12.Qe2 Bb7;  13.Rad1 h6; 14.Rfe1 Rfd8;  15.Bb1,  
      15...Bf8;  "~"  (Maybe - "+/=")  {Diagram?}  
       ... "with chances for both sides."  - GM Nick de Firmian.   
      V. Ivanov - Zagrebelny;  Moscow, 1995.  

     [ See MCO - 14;  page # 350, column # 4, and note # (u.). ]  ]   

Both sides continue to develop in - pretty much - a normal manner, here.  
6.Be2 e6;  7.h3!?,  {Diagram?} 
A nice move, kicking away the pesky Bishop. 
(Some pundits ...  prematurely ...  said the way White handled this opening 
 completely refuted Black's approach to this opening. STUPID!) 

But, in Game Three of this match, the computer would eliminate this move, 
and do without h3 entirely. 

     [  = 7.Na3!? cxd4;  8.Nb5!, "~" {Diagram?} 
            The position is not at all clear. 

***

        This SAME opening was repeated in game # 3 of this match.  
        There the continuation was: = 7.0-0 Nc6;  8.Be3!? cxd4;  9.cxd4,  
        9...Bb4!?; {Diagram?}  This looks risky, but Kasparov is MR. CHESS! 
          (Playable had to be: 9...Be7!?, "=")    10.a3 Ba5;  11.Nc3 Qd6;  
        12.Ne5!?, "~" {Diagram?}   and, (in my opinion); Black was a little 
        better coming out of the opening.   

        Deep Blue  - GM G. Kasparov;  Man vs. Machine/ACM Match.  
        Game Three (#3) Philadelphia, PA; (USA) 1996. ]  

7...Bh5;  8.0-0,  {Diagram?}   
It makes perfect sense to castle here. But this was a relatively new 
continuation, and had not really been worked out to any great depth. 

     [  Kasparov has played this whole line before:  8.Be3!? cxd4;  9.cxd4,  
        9...Bb4+; 10.Nc3 0-0; 11.0-0 Qa5!;  12.Qb3!? Nc6;  13.a3 Bxc3;   
        14.bxc3 Nd5;  15.Rac1!?, "="  {Diagram?}  
        The game is equal, or at best - White has a very TINY edge here.  

        GM V. Kramnik - GM G. Kasparov;  
        Match Game, (Speed) Paris, FRA; 1994. ]  

8...Nc6;  9.Be3 cxd4;  10.cxd4 Bb4;  (TN?)  {Diagram?}  
"A new move," according to the book by Michael Khodarkovsky. 

     [  The continuation of:  10...Be7;  11.Nc3, Qa5;  12.a3,  {Diagram?} 
         is slightly better for White according to  GM L. Shamkovich.  ]  

11.a3 Ba5;  12.Nc3 Qd6;  
13.Nb5 Qe7!?;   {See the diagram just below.}  
Several annotators roundly condemn this move ...  
yet I am certain it is fully playable.   

 '?!' - GM Nick de Firmian. (MCO)  

Black definitely could play ...Qd5!;  when Nc3, Qd6!?; Nb5, Qd5!; 
is a simple repetition of moves. And if in answer to ...Qd5!;  White 
plays Bc4!?, Black can respond with the sharp move, ...Bxf3!  

 But to tell the truth,  ...Qe7!?  is probably Black's third or fourth best move here.  

***

     [  Much better than the text move was:  >= 13...Qd5!; "~" {Diagram?}  
         - MCO.  (The book, "Modern Chess Openings." 14th Edition.) 

        Black might even be able to play the line: 13...Qd7!?;  14.Ne5 Bxe2;  
        15.Qxe2 Qd8; "~"  {Diagram?}  when White may have a very small 
        advantage, but the second player's position is certainly very playable.  ]  

***

Black will now play a whole series of second-best moves. 
This run of bad moves is what endangers Black's game, much 
more so than the slightly inaccurate ...Qe7.   {A.J.G.}  

14.Ne5 Bxe2;  15.Qxe2 0-0;  16.Rac1 Rac8!?;  {Diagram?} 
This looks logical ... but it also might be a little less than best. 

     [ Maybe better was: >= 16...a6!; "~" {Diagram?} ]

17.Bg5!, "+/="  17...Bb6!?;  {Diagram?}  
Once again ... a second-best move for Black. (Maybe.)  

It looks like Black should simply centralize his last Rook. 

     [  Better than the game would seem to be: >= 17...Rfd8!;  18.Bxf6 gxf6; 
        19.Nc4 Bc7;  20.Rfd1, "+/="  {Diagram?}   White has a VERY small 
         edge, but Black's position looks very defensible. ]  

18.Bxf6!,  {Diagram?} 
GM Daniel King gives this move an exclam, and probably rightfully so ... 
other moves give White no advantage at all. 

     [ 18.Qe3, Rfd8;  "~"  (Maybe equal, or "=".)  ]   

18...gxf6;  {Diagram?}  
This is virtually forced. 

     [ 18...Qxf6?!; ('?') 19.Nd7, "+/"  {Diagram?}  forks a Queen and a Rook.  ] 

19.Nc4!,  {Diagram?}  
"An excellent move, which Kasparov had apparently overlooked. 
  He cannot now capture on d4, since, ultimately, Qg4+ will win a piece." 
  - GM R. Keene  and  Byron Jacobs 

GM Daniel King also gives this move an exclamation point here. 

     [  Maybe not as accurate was: 
        19.Qg4+!? Kh8;  20.Nc4 f5; "~" {Diagram?} 
        The position is unclear ...  perhaps a tiny edge for White. ]  

19...Rfd8?!;   (Maybe - '?')   {Diagram?}  
Black allows himself doubled and isolated pawns on the Q-side. This is a 
bad move, and virtually all the computer programs notice this right away. 
(The evaluations of the position change almost immediately.) 

It would seem that Black had to play the move: ...Bc7; which looks 
very close to being forced. 

     [ It seems Black had to play: 19...Bc7[];  20.Rfd1!? f5!;  21.Nxc7!?,   
        21...Rxc7;  22.Qf3, "+/="  {Diagram?}   when White has a very solid 
        edge ... but a forced win is nowhere in sight!!  ]  

***

After the very inaccurate move played by Black on his last turn, White's 
edge and pressure grows with every move. Soon Black feels compelled 
to shed a few pawns in order to keep from losing. 

(GM Danny King gives White's next move an exclam, i.e. 20.Nxb6!) 
20.Nxb6 axb6;  21.Rfd1!?,  {Diagram?}  
This appears good, but Rc4 was also worth a look. 

     [ 21.Rc4!? ]   

21...f5; {Diagram?} 
Black must have felt it necessary to keep White's pieces ... 
especially the Queen ... off the e4-square.   

***

White now will break up Black's pawns here. (using d4-d5) 
Whilst many authors have rained exclams on White's play, 
I prefer to adopt a more reserved judgment.  

(GM Danny King gives 22.Qe3! an exclamation mark here.) 
22.Qe3!? Qf6!?;  23.d5!?,  (Maybe/probably - '!')  {Diagram?}  
White slams ahead ... and breaks up Black's pawn formation. 
('!' - GM Danny King.) 
('!' - GM Garry Kasparov  and  M. Khodarkovsky.) 

Apparently this temporary Pawn sacrifice took Kasparov by 
complete surprise."  - GM Danny King.

(Many annotators - including GM Kasparov himself - greatly praised and 
 and extolled this move.) 

     [ 23.Qg3+!?;  23.Rc4 ]  

23...Rxd5!?;  24.Rxd5 exd5;  {Diagram?}   
Black is temporarily a pawn ahead, but he will not be able to hold it.  

***

GM R. Keene  calls White's next move, " a very fine and patient play." 
(I am inclined to agree.) 

(GM Danny King gives 25.b3! an exclamation mark here.)   
25.b3!,  {Diagram?}  
"Perhaps the most intriguing move of the entire game. I can't imagine 
many humans finding this idea. Before capturing with the Queen on b6, 
White makes sure that the b2-pawn isn't left hanging. That is fair enough, 
but if I did have this cool idea, and had managed to calculate that Black 
was unable to hang on to his extra pawn, then I think I would have chosen 
25.b4 instead. It looks more  'natural'  for White to keep one of his pawns 
protected, rather than have both of them hanging ... though - and this is just 
a guess - there might have been an obscure variation in which the Knight on 
c6 comes to c4 via e5, causing difficulties. Incidentally 25...d4; would be 
met by 26.Nxd4."  - GM Danny King. 

I would simply add that in October of 2002,  Fritz 7.0  finds b3 here 
almost instantaneously. {A.J.G.} 

     [  Not as good was 25.Qxb6, ('?') 25...Qxb2. "="  Or 25.Rc3!?  ]  

25...Kh8!?;  (Maybe - '?!')  {Diagram?}  
This looks interesting, and could even be interpreted as a logical attempt 
to use the half-open g-file, to try and generate some play in this position. 

GM Raymond Keene gives this move a whole question mark, ('?') 
and says that this is the basically the losing move. 

 Even if I were to agree with GM Keene,   (Which I do NOT!);   to blame 
 Black's loss on one bad move - in a position where White already has 
 a VERY LARGE edge - is simply stupid ... and typical of some of Keene's  
 very shoddy analysis.  (And it is also like locking the barn door ...  
 AFTER the horse was already stolen!)    

GM Daniel King offers a completely different viewpoint here.

     [ GM R. Keene says the following continuation is a much more 
       tenacious defense than the actual game: >=  25...Rd8!;  26.Qxb6,  
       26...Rd7;  27.Re1, "+/="  {Diagram?}  White is clearly better, but 
       Black has a VERY solid defensive position.   

***

       Another possibility was the rather surprising line of: 
       25...Ne7;  26.Rxc8+, Nxc8;  27.Qe8+!?  {Diagram?} 
       This is not forced, but it does win material. (I am almost sure that 
       Deep Blue would have done this.) 

          27.Qg3!?+  ("+/=")     

       27...Kg7;  28.Qxc8, Qa1+;  29.Kh2, Qe5+;  30.g3, {Box.}  {Diag?}   
       This is forced, (for White.). 

          30.Kg1, Qa1+;  31.Kh2, Qe5+;  32.Kg1,  {Diagram?}   
          is an obvious draw by repetition.   

       30...Qe2!;  {Diagram?} 
       This hits both f2 and b5. 

       31.Qxf5, Qxb5;  32.b4, "+/="  {Diagram?} 
       White is a shade better, but somehow I doubt DB would 
       win this position from Garry.  ]   

26.Qxb6 Rg8;  27.Qc5!;   {See the diagram just below.}    
White's edge now grows almost with every move.  

     [ Not  27.Qxb7?? Qg5; ("-/+")  or  27.Kh1!? Rxg2!; "<=>" ]  

27...d4!?;  {Diagram?} 
Black saves his d-pawn, but he might have been much better off ... 
by simply sacrificing it. 

     [ >=  27...f4!; "<=>" ]

28.Nd6!,  {Diagram?}  
The best play, the energy in White's position increases greatly after this move. 

I must say that Black did not lose this game, nearly so much as White just won it.  

     [ 28.Rd1!?, "+/=" ]  

28...f4!?;  {Diagram?}   
Kasparov tries an attack against White's King here. 
(Its not like he had a choice. What can he do? Should he sit back and 
 wait for White to munch out on all his pawns, then run his connected, 
 passed-pawns on the Queens-side all the way up the board?)

Keene does make the good point here, that while a human opponent might 
have been {overly} impressed and intimidated by the coming attack on the 
K-side, Deep Blue has no worries here. (It does NOT get tired, nor will it fall 
prey to an anxiety attack. It simply performs its computations, and produces 
the move that its program requires of it. A reason for this is that the machine - 
when it is working perfectly - will do something a human will not. It will ALWAYS 
find the tactics and the threats, no matter how deeply they might be hidden.) 

29.Nxb7,  (Possibly - '!')  {Diagram?}  
"An extraordinary move. If a human had played this against Kasparov - 
  I would have said that he was very brave, or more likely, very foolish. 
  As it is a computer, particularly one like Deep Blue, then bravery and  
  foolishness don't come into it. It's just correct."   - GM Daniel King. 

     [ 29.Qd5!? ]   

29...Ne5;  {Diagram?} 
Getting the Knight over to the King-side, and setting a very obvious trap. 

30.Qd5,  {Diagram?}  
White correctly defends the light squares near his King. 

     [ 30.Qxd4?? Nf3+;  31.Kf1 Nxd4; "-/+" ]  

30...f3;  31.g3,  (a fun position)   {See the diagram just below.}    
No exclam here, this is practically forced.  

But I must admit, Black's coming attack looks VERY daunting. 

  "This is clearly the crux of the game. As so often in games against    
    human victims, Kasparov has sacrificed his entire Queen-side in   
    order to build up immense pressure against the enemy King."    
     - GM Raymond Keene  and  B. Jacobs.