GM Garry Kasparov - Fritz_X3D (C);  
Game One (# 1.)

   GM Garry Kasparov (2830) - FRITZ_X3D (2675)   
[D45]
  Human vs. Computer/VR/3D  
  Athletic Club, New York; NY/USA  (Rd. # 1),  11.11.2003  

[A.J. Goldsby I]

This is the first game of the match between Garry Kasparov and Fritz X3D ... 
the computer. (program) 

"Man vs. Machine" World Championship Match.  
( Garry Kasparov's opponent here is the latest version of Fritz, running on a 
  multi-processor machine. Further, there is NO real chess set ...  
  the entire duel is being fought in space!  
 {Or, more correctly, in virtual reality.} )   

********************************************************************************************************

1.Nf3,   {Diagram?}     
White plays a true Reti ... but you can bet that:  
A.)  Garry intends to transpose into some other line. 
B.)  He had a reason for choosing this particular move order!  

The Reti is one of the most flexible of all opening systems ... it is also the most 
transpositional. {It is possible to play literally dozens of different openings from 
this one move.}  Therefore, because of the extreme difficulty of this line - and the 
opening knowledge required to play it correctly - it could be one of the most difficult 
of all the different systems the first player could use to start a chess game. 

     [ More normal for Garry is the Opening:  1.d4,  {Diagram?}    
       as he used in the European Team (club) Championships. 
       (This event was played in Greece in August.) ]   

1...d5; 
A good move - which controls the center - but certainly not the only move. 

     [ Black could also play:  1...Nf6;  {Diagram?}  
       leading to a completely different type of game for the second player.  

       With the moves:   1...c5!?;  2.e4!?,  {Diagram?}   
       we are playing a  Sicilian Defense.     

           (And after 2.c4, we have instead the Symmetrical English.)  ]       

2.c4,  
This move hits the center and is the normal move for White here, especially 
if he wants to play a  'true'  Reti. 
 (2.d4,  2.g3,  and  2.b3,  are also possible in this position.)  

     [ With the move:  2.g3,  {Diagram?}  the first player offers to play 
       a kind of "King's Indian Defence," ... but with colors reversed!! ]  

2...c6; 
Generally, if your opponent attacks your foot soldiers in the center with a Pawn, 
you should ALSO defend with a Pawn. (To keep from losing time in the opening.) 
This move has the additional benefit of also {possibly} threatening ...d5xc4 for 
Black.) 

With this play, (1...P-QB3 or ...c6;); Black also can transpose into the main lines 
of a Slav Defense.  

*************************

     [ Another line that the second player can also use is the continuation:  
        2...e6;  3.b3!?,  {Diagram?}  
        This is one of the more interesting lines that White can use here ... 
        but it is certainly not the only one!  

           ( With the moves:  3.d4 dxc4;  4.e3 Nf6;  {Diagram?}      
             Play will most likely transpose into yet another main line ...     
             The Queen's Gambit ACCEPTED. )       

        3...Nf6;  {Diagram?}   
        Simple development is probably the best ... pushing the Black Pawn 
        to the c5-square is the most ambitious approach for the second 
        player in this position. 

        4.g3 Be7;  5.Bg2 0-0;  6.0-0 c6!?;  7.Bb2 Nbd7;  8.d3, "+/="  {Diagram?}  
        which Reti himself used in a tournament in the 1920's.  
        (White has a very small advantage.)  

        There are literally hundreds - if not thousands - of Master games in this line. 
        Probably the most recent example, with GM's involved - is the contest:  
        GM J. Horvath - T. Farkas;  The Hungarian Championship(s),  
        (tourn.) chT3, of02 (f), 2001.   

***************

        With the moves:  2...e6;  3.d4,  {Diagram?}     
        Heading back to the main lines of the Queen's Pawn Openings.

           ( Instead, with the moves:     
             3.g3 Nf6;  4.Bg2 c6; 5.b3 Be7;  6.0-0 0-0; 7.Bb2 Nbd7;   8.d3, "+/="  {D?}      
             we play this opening line the way that Reti actually believed it should be      
             played. )       

        3...Nf6;  4.Bg5 Nbd7;  5.Nc3 c6;  {Diagram?}     
        we reach the main lines of the normal variation of the line known 
        as,  "The Queen's Gambit Declined."  
        (This line is over 150 years old!)  ]   

*************************

3.d4, 
One of the most consistent moves, White occupies and controls the center. 
With this move, we could also transpose directly to a "Queen's Gambit Declined," 
if both players are so inclined.

      [ It was not too late to play the move:   3.g3;  {Diagram?}     
        trying to steer the game back into a 'true' Reti Opening, or 
        even the Catalan System.  (And 3.b3 is also fully playable.)  

*******

        After the moves - and the rather unusual move order - we 
        reach the following position:  
        3.b3!? h6!?;  4.Bb2 Bf5!;  5.d3 Nd7;  6.g3 Ngf6;  7.Bg2 e6;  
        8.0-0 Bc5;  "~"  {Diagram?}      
        This is the  "Lasker's Defense"  to the  Reti Opening ... 
         a line I used in tournaments for over 30 years. ]   

3...Nf6;  
One of the most consistent and reliable moves for Black. The 2nd player 
controls the center, develops a piece, and prepares castling.  
(These are 3 out the 4 basic principles of the opening.)  

*************************

     [ Black can also try:  3...e6!?;  {Diagram?}     
       in this particular position. 

       After the {somewhat} wild moves:  
       4.Nc3!? dxc4!?;  {Diagram?}    
       The most provocative  ... ... ... and also the most ambitious.  

           ( After the moves: 4...Nf6; 5.Bg5 Nbd7; 6.e3 Qa5;  {Diagram?}      
              we reach the system of defence to the Queen's Gambit - that was    
              first worked out in detail ... by the great H.N. Pillsbury.      

              {THE CAMBRIDGE SPRINGS Defense.} )     

       5.a4 Bb4!?;  {Diagram?}   
       Exploiting the 'hole' on b4.   

       6.e3 b5!?;  7.Bd2 a5;  8.axb5 Bxc3;  9.Bxc3 cxb5;  10.b3 Bb7;     
       11.bxc4 b4;  12.Bb2 Nf6;  13.Bd3!? Nbd7;  14.0-0 0-0;  15.Qc2,     
       15...Qc7;  16.e4, "+/="  {Diagram?} (White has a small advantage.)     
       we reach ...   "The Noteboom Variation."    

       This line is one of the sharpest and most difficult lines that the 
       second player can employ in the Q.G.D.  

       Sanchon Guirado - Bellon Lopez;  Barcelona, (ESP);  1989.  

       See the book:  "Play The NoteBoom,"  by  Mark Van der Terf  and 
       also  Teun Van der Vorm.  (c) 1996.  (Published by Cadogan Chess.)  
       Chapter # 8, page # 79.  ISBN:  # 1-85744-108-7 

*******     *******     *******

       Black could also try:  3...dxc4!?;  4.e3!? b5!?; ('?!')   5.a4! e6;    
       6.b3! Bb4+;  7.Bd2 Bxd2+;  8.Nfxd2! cxb3;  9.axb5, "+/="  {Diag?}      
       I think that White is at least a little better in this particular position. 
       (The books consider this position to be much better for White, but virtually all 
        of the "GM-versus-GM" games in the database have been draws.) 

       I would have to say the most hard-fought contest would have been the 
       following key struggle:  

       GM Boris Gulko - GM Ivan Sokolov;  International Chess Tournament    
       /Groningen/NED/1994/  Drawn in 54 moves. ]    

*************************

4.Nc3,  
White simply develops a piece. (It also controls the center.)  

This is the simplest ... it is also the best.

Not to be recommended was Bg5 here, as ...Ne4; might force White 
to move his Bishop a second time.

     [ By playing the move:   4.e3,  {Diagram?}   White immediately guards 
        the c-pawn ... but also allows the 2nd player the possibility of playing 
        the Knight leap. (...Ne4) ]   

4...e6;  
The simplest ... and also quite possibly the best move for Black. 
(The only real drawback for the 2nd player is now the light-squared  
Bishop is closed in behind the Pawn Structure.)  

     [ With the moves:  4...dxc4!?;  5.a4!?,  {Diagram?}  
        This is not the only move ... e3 AND e4 are played here.  
            (Even Bg5 might work as well.)    

        The move a4 was played by both Rubinstein and Capablanca. But 
         it was probably the Russians who analyzed this system in great detail. 
         In particular, there are several nice wins - which I saw in a book of his 
         best games - by Botvinnik; which made an indelible impression on me. 

            ( The sharpest line for White is the move:  5.e4!?,   {Diagram?}     
               but some modern GM's consider this a somewhat risky approach.       
               (See the excellent book: "The Slav For The Tournament Player,"      
                by  GM Glenn Flear.)  )       

        5...Bf5;  6.e3 e6;  7.Bxc4 Bb4;  8.0-0 Nbd7;  9.Qe2 0-0; 
        10.e4 Bg6;  11.Bd3 Bh5;  {Diagram?}   
        The end of the column.   

        12.e5!?,  {Diagram?}   
         This is sharp - whether or not it is best in not at all certain.  

            ( Also possible was:  12.Bf4!?, at least according to MCO. )     

        12...Nd5;  13.Ne4,  ('!')  {Diagram?}   
         Probably the most accurate approach for White here.  

            ( Possibly  13.Nxd5!?, "~"  {Diagram?}  is OK for White, although         
              the first player  ...  maintains a {very} "minimal edge," according       
              to the respected author of MCO. ( - GM Nick de Firmian.) )       

        13...Be7;  14.Ng3 Bg6;  15.Bxg6 fxg6;  16.a5 a6!?;  17.Ne4 h6;  
        18.Ne1 Rf5;  19.Nd3,  "+/="  {Diagram?}   
          ... and the position:  "slightly favors White, though the game is  
         strategically complicated."  - GM Nick de Firmian.   

        GM J. Ehlvest - GM D. Barua;   The (FIDE) World Team Champ.  
        (Commonly known as an Olympiad.) / Yerevan, RUS;  1996.  

        [ See MCO-14;  page # 477,  column # 58, & also note # (m.). ]  ]   

*************************

5.e3,  {See the diagram ... just below.}     
Probably the simplest ... and also the sanest move. 
(White guards the attacked unit at c4 and prepares to develop his King-side.)  

{The main lines of this opening feature White playing Bd3 and then a quick  
  castling on the King-side.}  

***************

***************

     [ The sharpest lines for White here probably arise after the following   
        continuation:  5.Bg5!? dxc4!?;  6.e4!? b5!?;  7.e5 h6;  8.Bh4!? g5!?;    
        9.Nxg5! hxg5;  10.Bxg5 Nbd7;  11.exf6,  "+/="  {Diagram?}    
        when White probably maintains a small but clear advantage.  
        [ See MCO-14;  page # 473, columns # 44 through # 47. ]  

        This is the very sharp lines of the system known as:  
         "The Botvinnik {and Geller!}  (ANTI-MERAN) Gambit."  

        I remember a very famous game that was played between Garry Kasparov 
        and Mikhail Tal in this (very wild and crazy) line. 
        {This game was analyzed in depth by many different chess magazines.} 

        I own several books on this line and I have also played it numerous 
        times in 5-minute chess with other Masters and Experts. But this line is 
        probably best left to postal chess ...  and the world's experts of this sharp  
        line. (GM Lev Polugaeyevky - before he passed away - was considered 
        perhaps the world's leading expert in this particular variation ...  
        especially from the White side.)    

        Currently the best opening book on this particular  (ultra-sharp)  line is:  
        "The Botvinnik {Gambit} Semi-Slav,"  by  IM Steffen Pedersen. 
        Copyright (c) 2000.  Published by GAMBIT Books. (UK)  
        ISBN:  # 1-901983-26-9  

        The most recent (GM) example I could find in this particular variant  
        was the contest:  GM Pedrag Nikolic - GM Rainier Knaak;  
        European Club Championship;  Rethymnon, Greece;  2003.  
        (An almost unbelievable contest ...  that was drawn ...  
         in just under 50 moves.)  ]   

5...Nbd7;  
Once again ... this is probably the best method. Just simply develop ...  
and most problems will take care of themselves.  

Lasker said,  "Knights BEFORE the Bishops,"  ...  and here is a very 
good practical example of the correctness of this axiom.

      [ Currently - - - the ultra-sharp continuation of:  5...a6!?;  "~"  {Diagram?}  
         is VERY much in vogue. 

***

       The very safe and sane move of: 
        5...Be7;  {Diagram?}   is OK - developing the Bishop, and also getting 
        ready to castle - but isn't a very testing variation for the first player to meet. 
        (White gets a solid edge with simple and straight-forward developing moves.) ]  

6.Qc2,  {See the diagram just below.}    
A VERY old move ... that is currently all the rage again.
 (See Bareev's win over GM Ivanchuk from the FIDE Grand Prix, 2002.)    

***************

***************

This play, (Qc2 on move six);   was first tried   in the contest:  
O. Chajes - I. Kashdan;  New York Championships, (Rice CC)  
New York, NY;  (USA)  1924.  
(As far as I can tell from various CD-ROM and on-line databases.)  

I am sure the idea of Qc2 is to simply play positional chess ... and slowly 
increase White's small edge in space - in this position. (I think the other 
reason GM's use this move is to avoid the over-analyzed main lines of 
this opening.)   

According to the Informant DB ...  this position has now occurred literally 
thousands of times at the Master level!  (and above)    

*************************

     [ The main line occurs after the following continuation:  
         6.Bd3 dxc4;  7.Bxc4 b5;  8.Bd3 a6;  9.e4 c5!?;    
         10.e5!? cxd4!?;   11.Nxb5!,  "+/="  {Diagram?}       
         and White seems to maintain at least a slight edge.  
          (The  'Semi-Slav,'  Meran Variation.)    

         I stop here because the lines become very complicated, and it 
         is not really clear which of Black's responses is the very best.  

         [ See MCO-14; beginning on page  # 457; 
           columns # 1 all the way through column # 30. ]   

         See also the excellent book:  "The Complete Semi-Slav,"  
         by  IM Peter Wells.  Copyright (c) 1994.  
         (Published by Henry Holt Books.)  

         On a historical note, the move Bd3 on move six is very old. 
         The first strong master to use this move was probably  ... 
         H.N. Pillsbury in the late 1800's. ]   

*************************

6...Bd6;  
A common position in the Slav.  

This is  <also>  the same position as occurred in game one of Garry's match 
versus DEEP_Junior. (C) 
(There Garry scored a very nice and a very convincing victory.)  

---> BUT! Although here we probably used a different move order, the end 
result is exactly the same. In other words this is the exact same position
after six moves of Game One of Garry's Match versus DEEP-Junior.
(There the move order was  1.d4, d5.)  

Black's last move is VERY good. Black controls the center, (e5 & c5); develops 
a piece and prepares King-side castling. The second player is also prepared for 
{an eventual} freeing idea/break of ...d5xc4; and then ...e5! If the second player is 
able to successfully execute this freeing break - most of Black's problems will 
be completely solved.  

     [ Also good is:  6...Be7;  although the text (...Bd6) is sharper and more ambitious. ]   

Garry now continues with the same Pawn sacrifice that he (also) offered 
in game one of his match vs. the "Deep_Junior" chess program.  
7.g4!?,  
This offers a (possible) Pawn sacrifice that Black should simply decline.
  (See my  (See my  analysis  of the Kasparov - Deep-Junior game for more details.)     

Of course this is EXACTLY the same move that was played in ...   Game One   
 ...  of the  Kasparov - DEEP_JUNIOR (C)  Match.  

     [ Also ... VERY playable was the simple, but effective, move of:  
        >/=  7.Bd3,  {Diagram?}  followed by very rapid development.   

        After the further moves:  
        7...0-0;  8.0-0 dxc4;  9.Bxc4, "+/="  {Diagram?}   
        a very common position in the opening has been reached. 
        The literally hundreds of master-level games confirm the 
         judgment that White is probably slightly better.  

         [ See MCO-14, page # 468, column # 33, and all applicable 
           notes for this line. ] ]   

7...Bb4; ('!')  {Diagram?}      
This is probably the best move. 

It is also the main line, and I am tempted to award the machine an exclam for 
avoiding the silly and inferior garbage that occurred in the afore-mentioned game 
against the other program. (Deep Junior.)  

***************

      [ One on-line commentator suggested that Black should play:   
        7...b6;  {Diagram?}   but this strikes me as a little artificial. 
        (To say the least!)  

        The other contest saw the moves:  
         7...dxc4!?;  8.Bxc4 b6!?; ('?!')   9.e4!, "+/="  {Diagram?}   
          ... Garry already had a clear and very solid advantage ...  
        and went to score a brilliant and smashing victory.   

        GM G. Kasparov - DEEP_Junior (C);  
        "Man vs. Machine" Challenge Match; (Game # 1);  
         New York, NY;  2003. ]   

***************

8.Bd2 Qe7;  9.Rg1!?,  {See the diagram just below.}      
White appears to have a change of heart, but since Garry played 
this move SO quickly, I  must  believe that this was part-and-parcel of 
Kasparov's pre-game preparation.  

***************

***************

This move was first played back in 1995, but it did not meet with success 
in its initial outing. The computer(s) mostly consider White to be solidly 
better. Garry must have noticed this, and decided Rg1 was a big improvement 
over the main lines.  

{The main drawback to this move is that White can no longer castle on 
 the King's side.} 

*************************

       [ White could also try:  
          9.g5 Bxc3;  10.Bxc3 Ne4;  11.Rg1 b6;  12.0-0-0 Bb7;  {Diagram?}   
          but  GM Nick de Firmian  considers this position to be UNCLEAR.  

          [ See MCO-14,  page # 468, col.'s 33-36, and note # (j.). ];  

**************

          Several books give the following line ... as the main line:  
          9.a3!? Bxc3;  10.Bxc3 b6!?;  11.Bd3 Ba6;  12.Qa4!?,  {Diagram?}  
          GM de Firmian condemns this as inferior ('?!'), but does not inform  
          us what move is superior.  

              ( Maybe  >/=  12.Bb4!?, ('!')  instead? )       

          12...dxc4!;  13.Qxa6 cxd3;  14.Qxd3 0-0;  15.g5 Nd5;  
          16.Bd2!?,  {Diagram?}    
           This looks too routine.   

              ( Maybe better is: >/=  16.e4! Nf4;  17.Qf1 f5;  18.e5 Nd5;        
                19.Qc4! c5;  20.0-0, "~"  {Diagram?}       
                 and White seems to have survived the worst of his problems.      
                 {A.J.G.} )     

          16...f5!;  "=/+"  {Diagram?}   
           GM Nick de Firmian considers Black to be CLEARLY better 
           here, ('/+'); ...  and he may be right.   

           [ See MCO-14,  page # 468; columns 33 - 36 ...   
             and {also} (mainly) note # (j.),  Part (B.). ]   

          GM B. Gelfand - GM V. Kramnik;  EU Cup / Berlin, Germany; 1996. ]    

*************************

9...Bxc3;  10.Bxc3 Ne4;   {See the diagram, just below here.}        
At this point - - - I am pretty sure we are still  ... "in book." 
(Or at least I am sure we have NOT left the confines of established 
 Master practice.)  

***************

***************

11.0-0-0,  (TN?)  {Diagram?}      
This was all played very quickly ...  it must have been THOROUGHLY   
prepared ...  - IN ADVANCE!! - by BOTH sides for this contest.  

     [ Playable was:  11.g5;  {Diagram?}   and also   11.Bd3,  {Diagram?}     
        with approximately a level game. ]     

11...Qf6;  (TN?)  {See the diagram ... just below.}     
A big surprise by the computer.  

***************

***************

The positives of this try is that Black is attacking White's Knight on f3. 
{Which is currently unprotected.}

The main drawback is that this is the second time the Black Queen has moved 
in the opening  ... there is a definite loss of time with this method of playing. 

*******

Garry now sacrifices a Pawn for play. The next series of moves, (until White's 
17th move); appear close to being forced.  
12.Be2! Nxf2;  13.Rdf1 Ne4;  14.Bb4! c5!;  
Unless Black plays this move, (and blocks the a3-f8 diagonal); he will never be 
able to castle and get his King to safety.  

***************

     [ Black could have also played a line like the following:  
       14...Qd8;  15.Bd3 Ng5;  {Diagram?}   
       This could be best ... but nothing is set in cement here.  

*****

            a).  Black maybe could play:  
                  15...a5!?;  16.Be1!?, "~"  {Diagram?} 
                   with an unclear position.   

                        (Possibly  16.Ba3,  here?)     

            b).  But definitely not:  15...f5?;  16.gxf5 exf5; 
                  17.Rxg7, '±'  {Diag?}   and White is very clearly better.  

*****

       16.Nxg5!? Qxg5; 17.Qf2!? Qf6;  18.Qg3 Qd8;  19.g5, "comp"  {Diag?}   
        (but) White has  "comp"  here ...  as Black's King is stuck in the center.  
        (The second player is also {obviously} very behind the first player in 
          development here.)   {Analysis.}  ]   

***************

15.cxd5!,  
This appears to be the best capture for White.  

     [ White can also play the other capture in this position, 
       but after the following moves:  
       15.dxc5!? Qe7;  16.Rd1!? b6;  17.cxd5 exd5;   
       18.Rxd5!? bxc5!; "~"  {Diagram?} 
        I think that Black is OK here.  (At least equal, "=" maybe better.) ]   

15...exd5[]; 
I think that this is pretty much forced for Black in this position.  

     [ The other capture is MUCH worse for the second player, viz: 
        </=  15...cxb4?;  16.Qxe4 0-0;  17.Bd3 g6;  18.dxe6 Qxe6;   
         19.Qxe6 fxe6;  20.h4, '±'  ('+/')  {Diagram?}    
         and White is very clearly much better in this position.  ]   

16.dxc5 Qe7; 
This is also {probably} the best  move for Black here.  

     [ Black could also try 0-0 here, but that is not really clear.  
       16...0-0!?;  17.Nd2!?,   {Diagram?}   
       This is good, but maybe White should play c6! here as 
        in the actual game.  

          ( Probably  >/=  17.c6!,  "+/="  is better. )     

        17...Qh6;  18.Bd3 Nxd2;  19.g5! Qh3;  20.Qxd2 Ne5; "~"  {Diag?}   
         Black could be OK  (maybe "=/+")  in this position. ]    

17.Nd4!,  
This is probably best.  

     [ After the moves:  
       17.Nd2 Nexc5;  18.Bb5 b6;  19.Bc6 Rb8;  20.Bxd5,       
       20...0-0; "~" {Diagram?}    the position is probably unclear. ]  

*******

17...0-0; ('!')  
This could be the best - some of the alternatives here for Black were 
distinctly UN-appealing.  

*************************

       [   Variation # 17B1.)    
          After the following continuation:  
          </=  17...Nexc5?;  18.Nf5 Qf8;  19.Rd1! g6;  20.Rxd5 gxf5;  21.Bxc5 Nxc5; 
          22.Rxc5! Bd7;  23.Bb5! Bxb5;  24.Re5+! Kd8;  25.Rxb5 Rc8;  26.Rd1+ Ke7;   
          27.Rc5 b6;  28.Rc7+ Rxc7;  29.Qxc7+ Kf6;  30.Qc6+,  ("+/-")  {Diagram?}      
          White should be winning. 

***************

           Variation # 17B2.)       
          After the moves:  </=   17...a6?!;  18.Nf5 Qf8;  19.Bf3 Nexc5;  {Diagram?}   
          This could be forced.  

               ( Not   19...Ndxc5??;  20.Bxe4, "+/-" )      

            20.Bxd5,   '±'  (Maybe "+/-")    {Diagram?}       White is clearly better. 

***************

           Variation # 17B3.)      
          The following line is inferior:  </=  17...Ne5?!;  18.Bb5+ Kf8;  {Diagram?}   
           (and) Black has lost his castling privileges.   ("+/=")     

               ( Even worse is: </= 18...Bd7?; 19.Nf5,  '±'  {Diagram?}   
                 and White is close to winning. )      

***************

           Variation # 17B4.)      
          After the following line:  </=  17...Ndxc5!?;  ('?!')   18.Bb5+ Kf8;   
          19.Kb1!,  "/\"  {Diagram?}     White is clearly a little better ...    
           and Black has difficulty in properly organizing his position.  ]  

  *************************  

The next few moves appear to be nearly forced and/or are best.  
18.Nf5 Qe5;  19.c6,  
Winning material.  

       [ Possible was:  19.Bd3!?,  with very dynamic play. ]   

The next couple of plays are also forced ... or best.  
19...bxc6;  20.Bxf8 Kxf8!; ('!!')  {See the diagram - just below.}       
A little bit of a shock, all the commentators were expecting Black to capture 
instead with the Knight here.  

***************

***************

Most GM's would have avoided this move on general principle. Fritz, with its 
ability to look at OVER four million positions-per-second, the program simply 
looked at all the lines and decided that its King was not in any real danger.   

     [ Maybe  20...Nxf8!?;  {Diag?}   instead? ]    

21.Ng3,   
Kasparov decides to rid himself of the Knight on e4.  

***************

     [ The computer ... and several on-line commentators ... suggested  
        the move:  21.h4!?,  {Diagram?}   maybe with the idea of advancing  
        on Black's King position ... and maybe creating some weaknesses.  

********

        A move like Qxc6!? could much too risky for White ...  
        but this is not really clear:   

        21.Qxc6!? Rb8;  22.Qc2 Ndc5;  23.Nd4 Bd7;  "<=>"  {Diagram?}   
        and I would say Black is OK here, he has good play for the material 
        he has sacrificed. ]    

***************

21...Ndc5;  22.Nxe4 Nxe4;  23.Bd3 Be6;  24.Bxe4!?,  {See the diagram below.}    
Garry played this rather quickly ...  but I am not at all convinced that this  
is the absolute best move here.  

***************

***************

{Opening the a2-g8 diagonal for the Black Bishop on e7 greatly strengthens 
  Black's position.}  

       [ I liked the move: >/=  24.Rf4!,  "+/="  {Diagram?}   for White at this point. ]    

24...dxe4;  25.Rf4 Bd5;  
Black's Bishop is a tower of strength on the d5-square ...  
and is almost impossible for White to remove by force.   

Many of the commentators were saying here that a kind of dynamic 
equilibrium has been reached.  

26.Qc5+ Kg8;  27.Rgf1,  
With the obvious threat of capture on Black's f7-square. 
(The Black Bishop would be unable to capture in the current position ... 
  because of the lateral pin ---> the BQ would be hanging!) 

     [ Possible was:  27.Rg2!?, "+/="  {Diagram?}       
        (and)  -  White is slightly better.  ]   

27...Rb8;  28.R1f2 Qc7;  {See the diagram just below.}     
This is a very important position.  

***************

***************

Now Yaz says Black should think about safe-guarding his a-pawn ...  
and he might be right. Instead Kasparov plays aggressively. 
29.Rc2,  
An aggressive move ... but one that also protects the White King
by keeping the pieces nearby.  

Of course White has the threat of QxB/d5!, PxQ;  RxQ/c7 ... winning.  

     [ Maybe better is: (>/=)   29.Kb1, "~"  {Diagram?}  
       and White maintains the superior long-term chances in this position. ]  

29...Qd7;  30.h4!?,  
Garry continues to play ...  "in his style,"  ...   which is to say, very actively.  

     [ The computer says it is better for White to play the move:   
        30.b3,  {Diag?}   but this creates a number of weaknesses as well. ]   

Thus far Garry has found nearly all of the really tricky moves, but now its is 
Fritz/X3D's turn to find the a fine tactical sequence.  

30...Qd8!;  
A good move - the main idea is to attack White's Pawns and also 
increase the scope of the Black Queen in this position.  

     [ I had expected:  30...Rb5; "~"  {Diagram?}    
        with good play for Black.  ]   

31.g5,  
Thematic - with Black having a light-squared Bishop, White should try and  
create weaknesses on the other color-square complex.  

     [ Also interesting is:  31.h5,  {Diagram?}     
        but Black retains fair chances. ]     

31...Bxa2!;  {See the diagram just below.}       
The computer cries: ENOUGH!  

***************

***************

This move not only re-establishes material equality, the loss of this pawn  
creates some distinct threats.  

     [ Perhaps Garry expected a continuation something like:  
        31...Qf8;  {Diagram?}    
        but after a move like:  32.h5!?, '±'   White is very clearly better. ]   

32.Rxe4, 
After a long think ... Garry finds the very best move.

***************************

     [ The continuation of:  </=  32.Qxa7? Ra8;  33.Qd4 Bd5; "=/+" {D?}     
        only creates open lines to the White King here.  

********

        And after the moves:   = 32.Rd2!? Qe8;  "~"  {Diagram?}     
        The computer retains the better practical chances. 
        (The tactics are going to be intense, the Queens are still on 
          the board, and Garry is running very short of time.) ]  

*************************

32...Qd3;  33.Rd4,  
This looks to be practically forced.  

     [ After the moves:  
       33.Qxc6 Qf1+;  34.Kd2 Rd8+;  35.Rd4 Rf8!; "~"  {Diagram?}   
       I would be very uncomfortable as White - with my naked King 
       stuck out right in the middle of the board. ]  

33...Qxe3+;  
Now the game will resolve itself into a draw by repetition.
(But there are still a few tricks left in the position.)  

34.Rcd2,  {See the diagram ... just below.}   
This is completely forced ... although many on-line {titled} commentators on 
ICC, and GM Yasser Seirawan did  NOT  immediately grasp this fact!!  

***************

***************

     [ A brilliant and shocking sequence would be the following moves:  
        </=  34.Rdd2?!, ('?')   34...Rxb2!!!;  {Diagram?}     
        A truly unexpected move.  

        And While this is best ... it may NOT be the only move for Black.   

          ( After the simple moves: </= 34...Qe1+!?;  35.Rd1 Qxh4;      
            36.Qe5 Rf8; "=/+"  {Diagram?}       
            I would much prefer to be Black ("/+")  in this position, as he has      
            a piece and two pawns for the exchange. )       

        35.Rxb2,  {Diagram?}    While this is very ugly ... it is probably forced.  

            ( NOT  35.Qxe3?? Rb1# .)     

        35...Qxc5+;  36.Rdc2 Qe3+!;  37.Rd2 Qc3+!;   38.Kd1,  {Diagram?}    
        This is pretty much forced.  

            ( </= 38.Rdc2?? Qe1#.  Or </= 38.Rbc2?? Qa1#. )     

        38...Bb3+;  39.Ke2 h6;  ("-/+")  {Diagram?}   
        with an easy win for Black. ]   

*************************

The next few moves are all pretty much forced for both sides.  
34...Qe1+;  35.Rd1 Qe3+;  36.R1d2 Qg1+;  37.Rd1,   {Final Diagram, below.}     
Here Garry talked to the arbiter  ...  the parties agreed to split the point.  

A truly great draw ... one I did not think would be exceeded for a number of 
years. Both sides played very well, attack and defense really canceled each 
other out.  

But with hindsight, after winning the exchange out of the opening; Garry must 
have been kicking himself for not winning this game. (One wonders if he spent 
a sleepless night trying to find all the improvements ... and if this somehow did 
not affect his play in Game Two - which he lost.)