I have taken my time annotating (& formatting) this game.   

My goal here was NOT speed, but accuracy. Many of the jobs of annotation contained gross inaccuracies or oversights. I also recorded all of the games coverage on TV, and I wanted to watch those tapes repeatedly in order to do the best possible job. (Although the television coverage of games 1-3 were interrupted ... something I have no control over. Only the television coverage of Game Four was complete.) But I am thankful I was able to record all the available TV coverage that was available. 

(I also wanted to give - at least in a small way - a flavor of what it must have been like to watch these games on TV. Where appropriate, I have 
  quoted the ESPN Team of commentators.  See  game four  for my breakdown of the ESPN Team.) 

Speed was not required, as I posted a short, briefly annotated version of this game within just a few hours of the completion of the first game. There were also other annotated versions of this game on the web, and I also provided links to those as well. 

 Wednesday;  February 04, 2004.  (And also May, 2004.)   

FRITZ_X3D (2675) - GM Garry Kasparov (2830)   
[C66]
  Human vs. Computer/VR/3D  
  Athletic Club, New York; NY/USA  (Rd. # 2),  13.11.2003  

[A.J. Goldsby I]

(Game No. Two {# 2.} of the Match)

The critical game of the match ... after a tenuous position, the computer did not follow up with the correct moves. Kasparov could simply move a piece back and forth and probably draw. Instead, after a blunder, the computer breaks through and wins. 

It was no surprise that the machine opens with e4 on the first move. The convention ... at the master-level ... is that this normally leads to a more tactical type of game, than most of White's other choices on the first move. 
(It is generally thought that one of the machine's strength's is - of course -  ... tactics!) 

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

The first five or so moves were made almost instantly.

 1.e4,  {Diagram?}     
Fritz wants an open game ... so that it can exploit its biggest gun. 
(The interplay of the pieces.) 

Personally, I would love to see Garry try to defend a King's Indian against the box. 

     [ It remains to be seen how Garry would have defended the opening: 
        1.d4,  (strategy)  {Diagram?}    
       in the second game. Would he have chosen the Q.G.A. as he did in game four? 
       Or was there some other opening prepared for the crucial second-round encounter ... 
       one we never got to see? ]   

 1...e5; ('!?')  {Diagram?}     
To me this is already something of a shock, I thought that - for sure! - 
Garry would play his beloved  Najdorf Sicilian  in response to "One - Pawn to e4." 

Perhaps Garry is avoiding this because it is so well known. Or perhaps Garry 
wanted to have the advantage of the FIRST surprise move in the opening! 

     [ Why not the move:   1...c5;  {Diagram?}  
        at this point? I think the answer is Garry is saving this move for a 
        situation when a win is absolutely necessary. (Or maybe he just had 
        something else in mind for this game.) ]   

 2.Nf3,  
The standard move for White at this point in the game. It controls the center, develops a piece, 
prepares King-side castling in just two more moves, and also attacks a very important and critical  
center-Pawn. (Forcing Black to address this threat.)   

     [  It would be interesting to see the machine play a gambit like    2.f4!?,  {Diag?}   
        I think that properly prepared with the very best lines, this would not be such 
        a bad choice. It also would regain the element of surprise. ]     

 2...Nc6; 
The most standard move at this point, at least at the "master-plus" level. 
It is good because it meets virtually all four of the opening principles.  

This is almost obligatory here. There is no way that Garry is going to play an extremely wild line, 
at least not this early in the match. 

Some examples of wildness are the Philidor Defense, (too passive); the Latvian Counter-gambit, 
(MUCH too wild to risk here ... also in some lines, memory plays as big a role as anything else.).  
The Petroff also has some very wild lines that Black can use, but a lot depends on which move - 
or moves - White chooses. Garry also hardly ever uses the Petroff that I know of.  Garry could 
have also played 2...d5; theoretically ... but I think there was almost no chance that Garry would 
play something really crazy.  

     [ The move: 2...d6!?; {Diagram?}  leads to the opening known as 
        the "Philidor's Defense." This opening mostly is very passive, and 
        I doubt if Garry would ever use this line ... unless he just wanted 
        to make a quick draw. (This opening also has some very wild lines, 
        but the likelihood of Kasparov employing any of these lines were 
        almost zero!) ]   

3.Bb5,  
The very venerable Ruy Lopez. This is one of the oldest known openings. 
It is also the best chance for White to get an advantage out of the opening 
after 1.e4, e5;  2.Nf3, Nc6.   

     [ Interesting was:  3.Nc3!?, {Diagram?}  when 3...Nc6; leads  
        to another very old opening, "The Four Knights." ]   

3...Nf6;   {See the diagram - just below.}      
SURPRISE!!  Will Garry use Kramnik's favorite line of the Berlin Defense? 
(White normally plays 4.0-0, and Black then plays ...Nxe4.) 

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

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

Garry's use of this opening is a radical departure from his previous use of opening systems. 
(To the best of my knowledge, Garry has never used the 'Berlin System' in a serious 
  tournament game.)  

     [ The 'main line' here is:   3...a6!?;  (Maybe - '!')  {Diagram?}     
        The MORPHY Defense, named after Paul Morphy. He was one 
        of the greatest players who ever lived and also was the first to show 
        some of the real strengths of the move, a3. 

        4.Ba4 Nf6;  5.0-0 Be7; {Diagram?}     
        This is known as:  "The Closed Defence," as one of my books calls it. 
        (The nature of the position stays closed - Black simply develops and 
         leaves White's e-pawn alone.) 

           ( Instead, after the move  5...Nxe4!?;  "~"   Maybe - "+/=" {Diag?}    
              we have:  "The Open Defence."  [ See MCO-14. ]           
              (Korchnoi plays/{played} this line at the highest levels, including      
                during a WC Match with Karpov. Anand also has played this variation       
                in a WCS match versus GM Garry Kasparov in New York in 1995.) )       

        (Returning to the main line of the Ruy Lopez.)  
        6.Re1 b5;  7.Bb3 d6;  8.c3 0-0;  9.h3!?,  ('!')   {Diagram?}   
        White plays h3 to prevent Ng4 ... and also any pins by Black's QB on 
        the White Knight on the f3-square. 

        White will play  d2-d4,  next ... and have a very good game. Literally 
         thousands of master games have been played from this position. 
         (Theory solidly favors White from this position.) 

        [ See the opening manual, MCO-14, beginning on page # 40. 
          (MCO has close to 60 pages of lines and analysis on this very 
            old and respected opening.) ]    

        To show how popular the Ruy Lopez is, I have over 50 books that deal 
         with this one opening system alone. (This is not counting general opening 
         manuals like MCO, ECO, NCO, etc.)  

        The most recent GM game I could find in my database was the contest:   
         GM Boris Gelfand - GM Giovanni Vescovi;    
         ICT / Bermuda GM (4) / 2004.   (The game was eventually drawn.)  ]     

The computer plays a slow line designed to avoid the loss of its vital center pawn.  
4.d3!?,  (ditto!)   
This is surely a line given to the computer by its team of programmers. 
(To avoid the main lines of the Berlin Defense, the opening line that has brought  
 GM Vladimir Kramnik so much success.) 

But in a way, this is already a big success for Garry. We are out of the main variations 
of the machine's opening book. Here the emphasis is already on general chess knowledge, 
and less on the machine's vast 'opening book' ... which has zillions of different lines in it. 

<< According to Alex Kure, the opening book expert for Team Fritz, they wanted to keep 
      the queens on the board ... >>  - Mig Greengard.  (ChessBase) 

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

     [ After the moves:  4.0-0 Nxe4;  5.d4 Nd6;  6.Bxc6 dxc6;  7.dxe5 Nf5;  
        8.Qxd8+ Kxd8;  "~"  {Diagram?}     
        Black has great defensive resources.  
        (According to a friend's database, which he maintains with a devotion to that approaching 
         a religion, this position has appeared about 2,700 times at the Master level.)  

        Garry himself was unable to break through Black's position in his 2000 (WC) {"Brain Games"} 
        Match with GM Vladimir Kramnik. 

        Also, Deep Fritz was unable to win as White against Kramnik in the ... "Brains in Bahrain" 
        Match in 2002. (Game # 1.) 

        A good {recent} example of this line was the following contest:  
        GM Peter Acs (2591) - GM Kiril Georgiev (2651);  
        The European Team (Club) Championships,  Plovdiv, BUL;  2003.  
        {Black eventually won the game.} 

*******

        One opening book gives the following continuation:   
        9.Nc3!? h6!?;  {Diagram?}   
        Preventing a White Knight or possibly a Bishop from coming to g5, but was this move really necessary? 

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

            ( Black could also try:  9...Bd7!?;  10.b3!? h6;  11.Bb2 Kc8;  12.Rad1 b6;  13.Ne2 c5;  14.c4 Bc6;     
              15.Nf4 Kb7;  16.Nd5 Ne7;  17.Rfe1, "+/="  {Diagram?}      
              and although White is, (or ... at least, appears to be); very solidly better in this position, Black was     
              able to ... {eventually} draw the game.      

              GM G. Kasparov - GM V. Kramnik;  "Brain Games" World Championship Match      
              (Game # 3); London, ENG; 10.2003. )         

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

       (We now return to the study of the known ... 'book' ... line.)   
        10.h3!? Be6;  11.g4 Ne7;  {Diagram?}    
        The end of the column.  

        12.Nd4 Bd7;  13.Bf4! c5!?;  14.Nde2 Kc8;  15.Rad1!?,  "+/="  {Diagram?}     
         "White must play very accurately to gain an edge against the under-rated Berlin Defense."    
           - GM Nick de Firmian 

        GM E. Geller - GM O. Romanishin;  45th URS Championship /Leningrad/U.S.S.R./1977.   

        [ See MCO-14; page # 45;  column # 8, and also notes numbers(s) # (a. - k.). ] ]   

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

The next few moves feature some rather simple, but logical and very straight-forward developing ideas.   
 4...d6;  {Diagram?}   
This is a safe and sane logical move, although it does close in Black's King-Bishop.   

     [ Paul Morphy would have probably played the move:   4...Bc5!?;  "<=>"  {Diagram?}   
       in this position. (But this could be slightly risky in my opinion.) ]     

 5.c3,    
Simple, safe developing moves.   

White is playing a very slow line, a highly positional method for handling this opening. 
(And since the computer was ripping these moves out nearly instantly, I think we can safely assume that 
 this all part of the machine's prepared book.)   

White's last move keeps Black's pieces off the d4-square, (And to a lesser extent, the b4-square as well.); and 
also prepares the {eventual} pawn push in the center of d3-d4.   

     [ An off-beat move, but one that was suggested by several amateurs, was:   5.c4!?,  {Diagram?}    
       I don't think this will be good for White, at least not in the long run. ]    

(Garry paused about one minute ... before playing his next move.) 
 5...g6;  ('!')   {Diagram?}     
Garry plays for a simple game where Black achieves a nice position simply by fianchettoing his KB. 
(I am pretty sure that this whole approach to this opening had probably been thought out by Garry - 
 and his team - in advance of this game.) 

After a little research, I found this idea was suggested by several different authors, i.e., GM Nick de Firmian. 
[ See MCO-14; pages 45-46, note # (u.). ] 

After weeks {months} of research, during which I perused almost all of the opening books in my library, it seems 
that the early ...g6; was (probably) originally an idea of Ratmir Kholmov. {He suggested this an answer to a slow 
line of the Ruy that was popular in the Soviet Union during the 1950's, but apparently he never had the chance to 
try this idea in an actual game.} 

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

In the next few plays ... both sides obviously want to continue their development ... and get castled. 
 6.0-0 Bg7;  7.Nbd2,  {See the diagram ... just below here.}    
While this move develops a Knight, it also blocks in White's QB in this position. 

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

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

Thus far play has developed slowly, but also in a highly positional manner. 
This is not inconsistent for this particular opening.  (The Ruy Lopez.) 

     [ Possible was:  7.h3!?, {Diagram?}   but to be honest, this is probably 
        a little too slow to be effective here. ]   

 7...0-0; 
Good, simple development. 

By castling, the Black King is put almost into the corner, where it is much safer than the middle of the 
board. (The Black Rook is also mobilized and brought closer to the center ... where it can play a more 
active role.)

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

While castling may not seem like a big deal to most players here, I have noticed a definite trend in GM chess lately. 
The fashion seems to be to 'push the envelope' ... ... ... 
and try to castle later and later ... in an effort to try and squeeze as much out of the opening as humanly possible. 

And on a similar note, I am currently (September, 2003; through February of 2004.); working my way through the 
very good book of 'short' games by GM John Nunn. ("101 Brilliant Chess Miniatures."  ISBN: # 1-901983-16-6. 
This book was published in 1999. The games in this book are all less than 25 moves - and all the players were rated 
2500 {FIDE} or greater.)  {In many of these games, it is extremely hard to pin down the losing move, even after 
many hours of study.} 

The above mentioned book, the winner castled (early) in 84 of the 101 total games, while the loser castled in 
only 27 of these contests. (See the introduction of the book.) 

Of course, most of the time when the loser was not able to castle, the victor often had to sacrifice material to try and 
prevent his opponent from getting the King to safety. But I think the moral is quite clear: FAILURE to CASTLE ... 
even at the exalted GM-level of play ... is not only risky; but more often than not leads to a complete failure!! 

     [ Another idea is to simply play a move like ...Bd7; followed the next move by ...a6; 
        in order to force White to make a decision about what to do with his Bishop on  
        the b5-square.  For example:   7...Bd7;  8.a4 a6!?;  9.Bc4 0-0;  10.a5, "+/="  {Diag?}     
        I think White is just a tiny bit better in this position.  

        GM V. Kupreichek (2515) - GM Alexander Aleksandrov (2535);   
        Minsk, USSR; / (Russia) / 1994.  (1/2 - 1/2; 41 moves.)  ]   

With his next move, White plays a Rook to the center of the board. This is both good, normal, and very commonplace 
in double-KP openings. (White also prepares a possible Nf1-g3; a very standard "Ruy Lopez" type maneuver.) 
 8.Re1 Re8!?;  (TN?)  
A simple strategy ... Rooks toward the center of the board. But Garry took over seven minutes to decide on this move.  

I think that Garry was really busy trying to decide on what type of general strategy that he should play, and wasn't really 
having any real difficulty at this point.  

One commentator on ICC said that this move was a novelty. (He must be right, I searched this position in the on-line 
ChessBase DataBase ... and this game was the ONLY match!)  

This move also was new or completely unexpected to the computer ...   
for the first time in this game the program did NOT respond instantly.  

     [ Also good was:   8...Bd7;  {Diagram?}     
        which I prefer over the move played in the actual game.   

        9.a4 a6;  10.Bc4 Be6;  11.a5 Bxc4;  12.dxc4!? Nh5;  "~"  {Diagram?}      
        The game was {eventually?} agreed drawn.   

        R. Ponomariov - A. Grischuk;   7th European Cup  (team tourn.)  
        Panormo, Crete / GRE / 2001.     (This game was drawn in only 12 moves.) 

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

       I definitely would have tried ... or at least been tempted to try, the move:   8...Ne7!?;  {Diag?}     
       with the idea that White's Bishop on b5 looks rather silly here. (But simply d4 may be an OK 
       response for White, even maintaining a small - but secure - advantage out of the opening.)  ]   

White now advances in the middle of the board ... gaining vital ground.
{The box took just a shade over 4 minutes ... to decide on this logical and very natural advance in the center.} 
 9.d4,  (Maybe - '!')  
The immediate advance in the center is probably best, although most of the time - in similar Ruy Lopez positions - 
White plays h3 BEFORE advancing to d4. 

     [ Interesting is:  9.h3!?, "~"  {Diag?}   with play analogous to other Lopez variations. ]     

 9...Bd7;  {See the diagram ... just below.}    
A simple move, developing a piece. Black also maintains a presence in the center and keeps the position 
more closed, (than open); in nature. (In addition, Garry has broken the pin on the a4-e8 diagonal.) 
{Which I am sure was part of the overall match strategy for Garry.} 

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

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

Seirawan also pointed out that Black is just slightly ahead in development at this point in the game. 

Garry took about three-and-a-half minutes on this move ... he was probably just checking the tactics ... 
or trying to remember all the stuff he looked at in preparation for this particular contest. 

     [ The exchange in the center:  9...exd4!?;  10.cxd4,  "+/="  {Diagram?}    
        was certainly playable for Black, but leads to a completely different 
        type of game. (White's semi-mobile pawn center could present a real 
        danger to Kasparov here.)  ]   

The box took about 10 minutes before deciding on its next move.  
 10.d5!?,  "+/="   
This move, while gaining a great deal of space, was something of a surprise for me. 
(It allows a position that it is more closed in nature, and <theoretically> more suitable 
 to the human player than a wide-open tactical melee.) 

I am sure Fritz made this move because after spending a good deal of time on this move, 
the program's evaluation functions simply informed the box that this was the play which 
gave it the greatest advantage.  

     [ I expected something like:  10.Bd3!?, "+/="  {Diagram?}    
        with a slight advantage for White.   

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

       Or even the move:  10.Qb3, "+/="  {Diagram?}    
       which could lead to a real tactical dog-fight for both sides. ]   

Garry took very little time before playing his response to d5.   
 10...Ne7;  
The only good move for Black, and a maneuver that would be familiar to Garry ...   
producing a type of ... "King's Indian Defense" position.   

GM Yasser Seirawan also pointed out that he played Garry many times  ...    
and this was exactly the type of position that he would definitely  ...  AVOID!! 

     [ The continuation of:  </= 10...Na5??;  {Diagram?}      
       "A Knight on the rim ... its chances are dim, its prospects are grim, and 
        the chances for success are ... indeed, very slim."  (An old chess proverb.) 

       11.Bxd7 Qxd7;  12.b4, ("+/-")  {Diagram?}    simply drops a piece for Black.  

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

       Also inadvisable for Black was the continuation of: 
       </= 10...Nb8?!; (?)  {Diagram?}   It is bad to move backwards! {here}   

       11.Qb3!, "+/="  {Diagram?}   and White has a very solid edge. ]   

 11.Bxd7,  ('!?')  
A straight-forward move, but something of a surprise to the panel of commentators at this point. 
(I also was surprised by this move, I expected something like a4, or maybe even Bf1 - to maintain 
 the tension on the chessboard.)   

     [ Maybe   (>/=) 11.Bf1!?, "+/="  {Diagram?}  was just slightly better? ]   

 11...Nxd7;  {Diagram?}     
I guess White could be a tiny bit better here ... but it is hard to be really sure. 

I prefer the move ...Qxd7! in this position. But I would also be the first to admit that Garry has a ton more 
experience here, and understands these types of positions more than I do. 

The position - while arising from a KP opening - has much in common with: "The King's Indian Defense." 
(This is an opening that computers have done very poorly in, at least in the past.) 

     [ I much prefer to play (the move of) ...Qxd7! 

       Just one sample line, (similar to the way the game was played); is the following continuation: 
       >/=  11...Qxd7!;  12.a4! Nh5!;  {Diagram?}      
       Obviously heading for f4.  

           ( </= 12...c5?!; 13.Nc4, "+/=" )     

       13.a5 a6;  14.b4 Nf4!?;  {Diagram?}  
        This is good, the move ...Bh6 was also worth some consideration.  

       15.h3!?, {Diagram?}    
        Played to prevent a later ...Qg4; {at some point} by Black.  

           ( Maybe better was: 15.Nc4!? "=" )     

       15...h6!?;  16.Qc2!? Rf8!;  17.c4!? f5;  "="  {Diagram?}     
        Black is at least equal in this position, it is possible the second player has already 
        stolen the initiative from White. 

       Original line of analysis, by - LM A.J. Goldsby I ]   

12.a4!?, (TN?)  (Probably - '!')   {See the diagram... just below.}    
Soltis said - in his excellent book, "Pawn Structure Chess," - that with a pawn on d5, White's play is on the Queen-side. 
(The panel of commentators also pointed this out. GM M. Ashley even gave the excellent analogy that you should 
  ... "always run with your blockers."  The pawn wedge on e4 and d5 seems to point at the Queen-side; so this is the 
  direction that White should head in.) 

This move is good and natural, it also gains some significant space on the Queen-side. Another very common motif in 
these types of positions is for White - if allowed - to run his a-pawn all the way to a6, creating more weaknesses in 
Black's Pawn Chain. 

  **************  

    --->   I searched about a dozen different databases for this particular position.   
  IT IS TRULY AMAZING  ... THAT IN AN OPENING AS WELL-KNOWN & EVEN SOMEWHAT 
  HACKNEYED AS THE RUY LOPEZ ... THAT AFTER ONLY TWELVE MOVES ... WE HAVE A 
  POSITION THAT HAS NEVER BEEN REACHED BEFORE, at least not at the master level!!!!!  

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

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

This unique position certainly deserves a diagram of the position. 

     [ More normal would be moves like:  12.Nf1!?, "+/="  {Diagram?}      
       (With a small advantage.)  

        Play could now proceed:  12...h6;  13.Ng3 Rf8;  14.a4!?, "+/="  (Maybe - '!')  {Diag?}      
        and White has a small advantage. 

       K. Langeweg - GM K. Spraggett;  ICT / Ibercaja Open / Zaragoza, ESP; 1992.    

          (That contest did not start with the KP, but instead opened with the move, 1.d4. 
            The first 11 moves of that game was:  1.d4, Nf6; 2.Nf3, d6; 3.Nbd2, g6; 4.e4, Bg7; 
            5.c3, 0-0; 6.Be2, Nc6;  7.0-0, Re8; 8.Bb5, Bd7; 9.Re1, e5;  10.d5, Ne7; 11.Bxd7, 
            11...Nxd7; and we have reached the same position after 11 moves - by both sides - 
            as in the Fritz_X3D - Kasparov game.);   

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

       Or White could even try the move:  12.Qc2!?, "+/="  {Diagram?}    
       with also a small advantage to White in this position. ]    

 12...h6;  
This move is {primarily} played to keep all of White's pieces off of the g5-square ... most notably a WN.
(Garry plans a long-term strategy of expanding on the King-side, perhaps followed by an attack on White's 
 King. In the meantime, the second player cannot allow White the very disruptive Knight maneuver of 
 N-from-f3-to-g5-to-e6.)   

     [ The continuation of:  </=  12...f5!?; ('?!')  13.a5! a6;  {Diagram?}   
        Black must stop the advance of this pawn. 

       14.Ng5! Nxd5!?;  15.Nf1! N5f6;  16.exf5 gxf5;  17.Ne3!, "+/="  {D?}   
        seems to clearly favor White. ]   

 13.a5!, 
A good move for White ... which gains space and threatens to play a6 ...   
obviously something Garry felt he could not allow. 

     [ Also good for White was:  13.b4!?, "+/="  {Diagram?}   
        or even   13.Nf1!?, "+/=" {Diagram?}      
        with a small plus for White in either case. ]   

 13...a6; 
This was virtually forced.  {Garry played this almost instantly.}    
(Black should not allow White to play his Pawn to this square, as the change in the Pawn Structure after 
 a4-a5-a6xb7 can only favor White.)  

White's center Pawns, esp. the one on the d5-square, makes finding the correct plan for Black a little 
difficult in this position.  

 14.b4 f5!?;  {Diagram?}     
Setting up a dynamic middle-game. White will pursue the opening of  lines on the Queen-side ... 
and Black must seek his chances on the King-side. 

(An idea borrowed from the "King's Indian Defence.") 

     [ Possible was:  14...g5!?; "~"  {Diagram?}    
        followed by an attack on the K-side. (Or at least the attempt.);  

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

       Also good for Black was the move:  14...Rf8;  "<=>"  {Diagram?}     
       preparing the ...f7-f5; lever. ]    

White - after the ...f5 lever put pressure on the center - desperately needs to shore up the center.   
 15.c4,  (Maybe - '!')  
"X3D Fritz is playing disturbingly good strategic pawn moves!" 
- A ChessBase commentator. 

This move supports the d-pawn, gains space ... and gives the computer the possibility of a later 
expansion (c4-c5) on the Queen-side. 

     [ After the moves:  15.exf5!? gxf5;  16.Nf1 Nf6; "~"  {Diagram?}   
        it is not clear who is better.   

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

       Maybe just:  15.Bb2!?,  "~"   ('+/=')  {Diagram?}   
        with maybe ('!') a tiny edge for White in this position.   

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

       Also playable for White was:   15.Rb1!?,  "+/="  {Diagram?}     
        with probably just a very small advantage for White here. ]    

 15...Nf6;   
The move of putting the Black Knight back on f6 is a common one in these types of positions.   
{If Garry is going to meet his goal of attacking on the King-side, he must begin to transfer all 
  of his pieces over to that sector of the board.} 

     [ Interesting was:  15...f4!?,  "--->"   {Diagram?}     
       when Black has an extremely strong King-side initiative in this position. ]   

 16.Bb2,  (dubious?)    
A strange looking move ... the Bishop on b2 appears to be ... "biting on granite" ... on the e5 square. 
(This move was almost universally condemned by the pundits, who said that Rb1!? or Qb3! was better.) 
Another aspect that one commentator pointed out is that normally in these types of positions, White needs 
the b-file for his heavy pieces. His last move just blocked that key line. 

On the other side of the coin, if the center ever opens up, White's Bishop will be positioned to control many 
key squares. 

     [ Better was:   >/=  16.Qb3!?, "+/="  ('!')  {Diagram?}      
        and White maintains a solid edge ... from this position. ]    

 16...Qd7!?;     {See the diagram ... just below.}         
Black looks to be fine here.  
{Garry took over 20 minutes to play this move.} 

 ************************* 

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

This move - (...Qd7;) - appears to be a natural one. (If Black wants to attack on the King-side, he will very naturally 
 want to bring his Queen to that sector, at least eventually.)  

The amazing thing about this play is that almost none of the  'expert'  commentators ... either on the Internet, or on the 
television ... actually predicted this move! 

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

     [ Many commentators on the Internet said that the move:   16...f4;  "~"  {Diagram?}   
        was indicated in this position.  (ESPN Commentator  ...  GM Yasser Seirawan.);    

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

       Another good move for Black was the try:   16...Rf8!?;  {Diagram?}      
        with pressure down the f-file.   

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

       A position suggested by one  TV  commentator was:   
        (</=)  16...c6!?;  17.c5!?;  "~"  {Diagram?}    
        ... "and the tension goes completely across the whole board ...   
         the chess board is in flames."   - GM Yasser Seirawan.  (ESPN)  

            ( I think better is:  (>/=) 17.Qb3!?,  "+/="  {Diagram?}   
               with some serious pressure for White in this position.)   ]     

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

White's next move is mysterious ... only Fritz knows what it really does to improve the computer's position.  
 17.Rb1!?,  ('?!')  (why?)   {See the diagram ... just below.}    
Possibly to put pressure down the b-file at some later point? 

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

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

All the commentators were puzzled ... and completely nonplussed by this particular move as well. 
(One titled player on ICC said this was a real BLUNDER ... and openly predicted that Kasparov would win easily!) 

The ONLY good thing (!) about this move ... as far as I could see ... 
was that the computer played this Rook sally almost instantly.  

     [  Variation # 17W1.)      
        I expected maybe:  17.Qc2!?,  "+/="  {Diagram?}   
        with very good play for White.  

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

        Variation # 17W2.)   
       Also possible was:   17.Ra3!?,    "+/="   {Diagram?}      
       with the simple idea of transferring this Rook (along the third rank); 
       over to the K-side ... for attack ... or for defense.  

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

        Variation # 17W3.)      
       One GM on the  "Play-Chess"  server ... suggested instead that White play:   
       17.Rc1!?,  "+/="  {Diagram?}    
        followed by c4-c5. (He claimed a small advantage for White.)  ]     

 17...g5!?;   {Diagram?}   
This move practically forces Garry to commit to a King-side attack. 

Garry took nearly 15 minutes on this seemingly obvious play. 
(I think Garry knew he wanted to attack, but there are several different ways for Black to implement his 
 assault from this particular position. he was probably trying to decide the best method for doing this.)

     [ Maybe the move:   >/=  17...Rf8!?; ('!')  {Diagram?}     
        instead? (I much prefer this.)  

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

       Also to be seriously considered was the move:   17...f4!?;  "/\"  {Diagram?}     
       followed by an all-out attack on the King-side. (This might have also been 
       better than the move played in the game.  - GM Alexander Khalifman.) 

       {GM Y. Seirawan mentioned this idea in his TV commentary many times as well.} ]   

Now White decides on the exchange of Pawns, viz. e4xf5. While this play was severely criticized by some 
of the pundits, it has the very desirable effects of opening the position, increasing the scope of White's pieces; 
and also weakening Black (somewhat) on the light-square complex. 
{The computer took only about three minutes to decide on this important move in this position!} 

When White captured on the f5-square with the Pawn, Black could have re-captured this foot soldier 
with his Knight ... but decides upon doing it with the Queen instead. Garry was probably thinking that 
exerting pressure on the half-open f-file ... is probably the more natural move for Black in that position.   
 18.exf5!? Qxf5!?;  19.Nf1!,    {See the diagram - just below.}      
An excellent move - the White steed moves from the rather passive square of d2 and heads for either g3 or e3. 
(The Knight's ultimate destination is the blockading/semi-outpost square on e4.) 

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

 ************************* 

The other positive thing about the re-deployment of the 'horsie,' is that this cavalry unit will gain time when it comes 
to g3 or e3 ... as it will be attacking the Black Queen. 

     [ Or  19.h3!?, "+/="  19...g4!?; "~"  {Diagram?}   
        and Black busts up White's King position. 
        (This was also suggested by one pundit on the ICC server.);   

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

       One 'expert' commentator also suggested the possibility of: 
       19.Rc1!? Ng6;  20.c5!? Nf4; "~"  {Diagram?}      
        ... "and Black has the initiative, and has good attacking chances on the King-side here."   
        - GM Yasser Seirawan. ]   

 19...Qh7!?;  (hmmm)   {See the diagram ... just below.}      
This is not a terrible move, but it looks like it is a tad artificial to me. 
(Perhaps Garry is hoping to be as brilliant as Nimzovich in his famous victory over Johner?) 

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

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

It seems that to me ... the most natural square ... in this position  ...  for the Queen is on either d7 or g6. 
(The only reason why Garry did not play ...Qg6 here is that he wanted to be able to play the maneuver of 
 ...Ng6-f4; and therefore needed the g6-square as the launching point for one of his Knights.)   
 {Kasparov took about 4.5 minutes to play this particular move here.}   

     [ GM Yasser Seirawan  liked the move:   >/=  19...Ng6; ('!')  {Diagram?}    
        with the idea of a later ...Nf4.  (This looks better to me than the move actually played.); 

       An extremely interesting line ... that took several hours to try and generate ... was:    
       19...g4!?;  20.Ng3 Qg6!?;  21.Nd2 h5!?;  22.Nge4 Rf8;  23.Nxf6+ Rxf6;  24.Ne4 Rf7;     
       25.Bc1! Raf8;  26.Rb2!?, "+/="  {Diagram?}    
       In this position, White looks to be very solidly better, although the game is far from over.  

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

        Another play in this position for Black would be:  19...Qg6!?;  {Diagram?}      
        with some attacking chances for the second player here. ]   

White now begins a policy of transferring and re-arranging his Knights.  
(This is wise, because on their current squares, the steeds cannot give the best defense to 
 Black's impending assault.)   

At about this move, we lost our continuous (TV) coverage of the game, as the sports network, ESPN; 
dumped our chess coverage in preference for a tennis (tournament), being played in Houston, TX.   
{The network promised regular updates on their "sports-news" channel ... but these were far and 
  very few in between.}   

 20.N3d2 Nf5;     
I don't know if there is anything now that is concretely better than this continuation for Black.   

     [ After the continuation of:   (</=)  20...Ng6!?; ('?!')  21.Ng3 Nf4;       
        22.Nde4, "+/="  (Probably - '±')  {Diagram?}    
        White is clearly better.  - GM Nigel Short ... for LCC. ]  

 21.Ne4 Nxe4;    
Black is a little cramped, and this exchange frees the second player just a little ...  
and also opens the f-file for Black's pieces.  

It also seems that this exchange must played sooner or later. 

     [ Also possible was: 21...Rf8!? ]   

 22.Rxe4 h5!?;  (Maybe - '?!')    {See the diagram -  just below.}     
Nothing about the position suggests that Black will succeed with this. 
(It looks like it might just create a weakness for Garry.) 

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

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

 {See the diagram just above.} 

Since Steinitz we have understood that all attacks must be based on some type of fundamental advantage in the position. (A weakness in the opponent's Pawn Structure, more space, or a much greater degree of mobility, at least in that sector of  the board. In my opinion, a great advantage can sometimes be a "three-to-one" ... {or greater} edge in the number of  
 pieces on one side of the board ... or the other.) 

Here Black enjoys NO edge of any kind that I can detect, and therefore his assault - while interesting - must {at least} 
be slightly speculative  ... ... ...  and therefore unsound. 

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

     [  Maybe  22...Re7!?; {Diagram?}   was better than the game?  

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

        Also possible was:   22...Qg6!?;  23.Qd3!, "+/="  {Diagram?}   
        but White is still {slightly} better in this position.  

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

        While it might be possible for Black to grab the "out-post" square here ...  
        I do not recommend it. Viz:   22...Nd4!?; ('?!')  23.Ng3, '±'  {Diag?}    
        White is clearly (much) better here.  

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

        Maybe also playable for Black was the rather simple idea of:   
        (>/=)  22...Rf8!?;  {Diagram?}   perhaps followed by Garry doubling his heavy 
        pieces on the half-open f-file? (The idea looks terribly obvious, and I see no instant, 
        easy refutation of this plan.) {A.J.G.} ]    

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

White now transfers his major pieces over to the King-side. This is a good idea - it improves White's defence, 
and it also improves the lines of communication {and control} within the first player's army. 
 23.Qd3!? Rf8;    
A quick up-date on ESPN news ... just enough time for us to see that Garry took just a little over two minutes 
to play this particular move. 

Black - apparently - is still massing on the King-side  ...  and is still looking for an attack in that sector. 

     [ One commentator ... on a European chess server  ...  suggested the continuation of:   
        </=  23...g4!?; ('?!')  24.Rc1!?,  {Diagram?}    I am not at all sure about this.  

           ( Maybe just: 24.Rbe1, "+/=" )      

       24...Re7!?;  25.c5!? Rf8;  26.Nd2?!, (Maybe - '?')  {Diagram?}      
       This is horrible. (I think that the idea is to transfer the Knight to the 
        c4-square, but this is simply too slow.)   

           ( The move >/=  26.c6!, {Diagram?} has to be better for White. )      

       26...Ref7;  "/\"  "~"  (Maybe - "=/+")  {Diagram?}    
       and Black has stolen the initiative from White. ]  

 24.Rbe1 Rf7;  25.R1e2!, 
I like this move ... the box solidly defends the f2-square, before proceeding with any other plans. 

I was surprised to learn {later} that others criticized this move. 

     [ Also very good for White was:   25.Nd2!?, "+/="  ('!')  {Diagram?}     
        with {perhaps} the idea of bringing the White Steed to the c4-square. ]   

 25...g4!?;  26.Qb3!?, ('?!')    {See the diagram - just below.}       
I do not believe this is the correct move for White. (And a possible waste of time.) 

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

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

White - perhaps several times in this particular game - builds up a {slight} advantage, and then just allows it to 
slip away through inaccurate moves.  

     [ Better is:  >/=  26.h3!, "+/="  {Diagram?}   with a possible King-side break-through for White.  
        - GM John Fedorowicz.  (On http://chess.fm.) ]  

 26...Raf8;  27.c5,  
While this seems a very natural advance or lever in this position, others wanted White to try 
something else in this position. 

     [ Maybe {also} good for White was:  >/=  27.b5!?, "~"  ("+/=")  {Diagram?}      
        with ... (maybe!) a slight edge to White in this particular position.   
        - GM Alexander Khalifman.  ]  

 27...Qg6;  28.cxd6!?,  (Maybe - '?!')    {See the diagram ... just below.}     
This is an extremely thoughtless exchange ... and completely blows any advantage that White was 
{slowly} managing to gain. 

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

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

White was on the verge of a break-through here, but throws it away with this one errant play. 

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

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

     [ Maybe simply:   28.Nd2!?, "+/="  {Diagram?}     
        with the idea of bringing the White Knight to c4?  

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

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

       Also worth some consideration here was the move Qc4!?  
       For example:  (>/=)  28.Qc4!?, Ne7[];  {Diagram?}     
       This could be nearly forced in this position.  

                                                   **********   

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

            ( Jennifer Shahade - in the pages of the Feb. 2004 {issue} of  'Chess Life'  - gives 
              some extremely poor and very shoddy analysis there. She recommends Rf6, giving  
              that move an exclam. But her work does NOT stand up to scrutiny!   

              For example:  </=  28...Rf6!?; ('?!' or '?')  29.c6!?;  {Diagram?}       
              A very strong try for White  ...  but NOT the only one.  

                  ( Very interesting were the massive complications that arise from the        
                     move of:  29.b5!?; "~"  {Diagram?} which might work out to the         
                     first player's advantage. )       

              29...Qf7?!; (Maybe - '?')  {Diagram?}      
              Simply a very, very bad move.  

***

                  ( Better was:  >/= 29...bxc6[];  30.dxc6+ Qf7!; "<=>"   {Diag?}        
                     (Black has good play here.)         

                    For if:  31.Qxa6?!,  {Diagram?}        
                    (This is probably too risky for White here.)       

                    then simply:  31...Ne7; "/+"  {Diagram?}  and Black is better.  )       

***

              30.cxb7 Ne7;  31.Qc2!,  {Diagram?}    
              This is a pretty good move. White simply covers f2 ... and prepares to ram home 
              the Q-side pawns in for a touchdown.  

                  ( Jennifer Shahade  only offers the totally idiotic line of:       
                     </=  31.b8Q?, ('??')  31...Rxb8;  32.Qxa6, "~"  {Diagram?}      
                     which is unclear at best. (And maybe - slightly - favors Black.) )      

              (Now we follow a line that was analyzed in some detail - on one  fairly popular  {chess}  website.)    
              31...Qe8;  {Diagram?}    Now this looks forced for Black.   

                  ( Or Black can try:   </= 31...Nxd5!?; ('?!')  {Diagram?}      
                     but now  32.b5! axb5!?;  33.a6, '±'  ("+/-")  {Diagram?}      
                     and  White is clearly better.  )        

              The next few moves look to be pretty much forced.   
              32.Ne3 Qb5;  33.Qc4 Qxc4;  {Diagram?}   This could be forced.   

*****

                  ( Probably slightly worse was:  </=  33...Qxb7?!; ('?')  {Diagram?}       
                     (Now Black falls prey to a combo.)      

                    34.Bxe5! dxe5?!;  {Diagram?}    
                    This is probably inferior.    

                        ( But after the moves:  >/=  34...R6f7;  35.Bxg7 Rxg7;       
                           36.Rc2, '±'  {Diagram?}  White is still much better. )        

                    35.d6+ R6f7!?;  36.dxe7 Re8;  37.Nf5,  ("+/-")  {Diagram?}      
                    and several different strong programs consider White to be completely 
                    winning in this particular position. )    

*****

              34.Nxc4!?,  {Diagram?}    While this is very good, maybe the natural-looking 
              move of taking with the Rook is even better.   

                  ( Maybe simply:  >/=  34.Rxc4!?, '±'  {Diagram?}        
                     when White is MUCH better. )         

              34...c6!?;  {Diagram?}   Black probably has to try this move ... or something like it.    

  *******  

                  ( Definitely not:   </=  34...Nxd5?!;  35.b5! Nb4!?;  36.Nxe5!, "+/-"  {Diagram?}   
                     --->  White is winning easily.         

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

                    Also worth a look was:  >/=  34...Rb8!?;  35.Nxe5! dxe5;  36.Bxe5 Nxd5[];         
                    37.Bxf6 Nxf6;  38.Rc4!? Rxb7;  39.Re6 Ra7;  "~"  {Diagram?}          
                    when a slightly unclear position has been reached ... which I think offers equal chances.        
                    (Black's Rook is too passive here to think seriously about winning the game.) )         

  *******  

              Intensive work with a strong computer program will clearly demonstrate that the next few  
              moves, (especially Black's!); are all (relatively) forced.    

              35.Nxe5!!,  {Diagram?}    A brilliant retort.   

                  ( But the very simple line of:  35.dxc6 Nxc6;  36.Nb6,  '±'  {Diagram?}       
                    could be just as effective.  ("+/-" ?)  )        

              35...dxe5;  36.Bxe5 cxd5!?;  37.Bxf6!? dxe4;  38.Bxe7 Rb8;    
              39.Rxe4 Rxb7;  40.f3,  '±'  {Diagram?}      
              Many strong programs assess this position as being simply a won game for White. 
              (The first player is a clear TWO pawns ahead in this position!)    

               { This line took a tremendous amount of time to verify. A.J.G. };  

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

                                                   ***********  

       29.Bc3! Bh6!; "~"  {Diagram?}      
       I would stop here, and evaluate this position as being VERY unclear, 
       but maybe {only} slightly favoring White.  

       These lines - in this note took DAYS of work ... and were verified on several different computers,  
       using programs like Fritz 8.0 and also ChessMaster 9000.   

       Shahade needs to buy a computer, and at least check her analysis! ]    

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

  *********************  

 28...cxd6; "="  (Maybe - "+/=")  {Diagram?}     
After getting a promising position ... White has frittered away any edge that he had. I am sure - with 
PROPER  play - that Black had nothing to fear from this position!   

White continues to probe, but without a blunder from Garry ... it would have all come to nothing. 
(Black's next few moves are forced, or help Garry to improve his overall position.)   

 29.b5!? axb5;  30.Qxb5 Bh6; ('!') 
Activating the Bishop on the (now) open line.   

     [ Possible was:  30...Rc8!? ]  

 31.Qb6,  (Maybe - '!?') 
A sneaky tactic in Kasparov's time pressure here, the box threatens to play a capture on e5 ... 
as Black's Queen is (currently) unprotected along White's sixth rank.