Site hosted by Build your free website today!

  Fritz_X3D (C) - GM Garry Kasparov; 

Game Four (# 4.)

"Man vs. Machine" World Championship Match
(The first chess match played in 3D/VIRTUAL REALITY)

 The Athletic Club/New York City, NY/USA; 2003. 

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

My goal here was NOT speed, but accuracy. I guarantee I have watched the tape for this game ... 
all the way through, from start to finish ...  at least 25 times now. 

The whole point of this game - for me - was to try and preserve the flavor of the on-line commentary. If you did not get to see these games live, or did not record them on the VCR, I wanted you to at least get a small taste of the excitement and 'the buzz' of what the guys on TV actually said during this game. 

I also  DID  NOT  want to simply "steal/borrow" from what the other writers and commentators said about this game. IN FACT ... I highly disagreed with much of what has been said and written about this game! So for me, annotating this game was about TRYING to correct all the stupid and silly things that have been written about this game. (Quality!) 

(Nov. 30th, 2003.) When I am done formatting this game, it will be close to 10 diagrams, and I will also have my own JS re-play page. But please be patient with me. Formatting pages is one of my LEAST liked duties or jobs - it is the most like pure work for me. (And easily one of the least enjoyable tasks that I do.) But I will continue to do this, because I demand near perfection in a web page! 

   Click   HERE   to see an explanation of the symbols I use to annotate a game.    

   Click   HERE   to see this game on a java-script replay page. (UN-annotated.) 

  Fritz_X3D (2675) - Garry Kasparov (2830)  
Human vs. Computer / Man vs. Machine W.C.M.

Match played in Virtual Reality/3D  ...  at the Athletic Club
New York, NY; {U.S.A.}  (Game # 4), 18.11.2003

[A.J. Goldsby I]

  kasp-vs-fritzx3d_g4-med.gif, 02 KB

One of the most critical games of the entire match. Since the score is currently tied, to lose would be a huge disaster for either party.

Garry ... normally ... would be out for blood, especially against a human opponent. Many predicted that Kaspy (Kasparov) would do exactly this - that he would go 'all out' for a victory. However, I counseled all my friends and such - in an email sent out the day  BEFORE  the game - that Garry would indeed try very hard. But he also would come prepared with a line - that in his own estimation, and not those of the many commentators - would allow him to keep (a)  "draw in hand." 

I have tried to concentrate here on the general ideas and verbiage of this game.
If you want to see a "theoretical" discussion, (all the applicable games of the line that was played in this game); - I suggest a good book on the Q.G.A. (You can also see the many excellent links and other jobs of annotations of this game that were done on the Internet.)


1.d4,  {Diagram?}    
Surprise number one. Seirawan said that the machine won in 2nd game with 1.e4. 
So why change?  

The simple answer was that game two was  NOT  clearly in machine's favor ...  
and the programmer's knew this.
(The program was relatively OK coming out of the opening, but I think it was clear 
 that Garry probably misplayed a very strong King-side attack!) Of course ...  
by now ... Garry had pretty thoroughly analyzed that encounter ... and may have 
come up with some ...  possibly very significant ...  improvements. 

Commentator P. Hoffman correctly noted that often times it is a very good 
strategy to keep your opponent a little in the dark - and guessing - as concerns 
the opening that you might actually choose. This also limits the enormous 
preparation that Garry is very justly famous for.  

     [ Game two saw the following continuation:  
        1.e4 e52.Nf3 Nc63.Bb5 Nf6!?4.d3 d65.c3 g6!6.0-0 Bg7;  
        7.Nbd2 0-08.Re1 Re8!?9.d4,  "+/="  {Diagram?}  
        and the computer had a small but fairly solid edge.  

        FRITZ_X3D - GM G. Kasparov;  "Man vs. Machine" W.C.M. (Game #2) 
        (A chess match played in VR.) / New York, NY; (USA) / Nov. 2003. ]  


Surprise number two. Garry plays the classical defense to White's QP opening.  

Although I was very shocked by this move, it does fall in line with what I had 
predicted for this game. (That Garry would play intelligently aggressive chess. 
He would play a line that had a lot of preparation behind it. He would definitely 
have something specific in mind ... possibly even a few TN's. He would try very 
hard for the win ... BUT! You could almost guarantee that he would also strive 
to keep the draw well in hand.)  

   Also - Garry's favorite move, at least when going for the win, has usually been     
   1...Nf6. But the programmer's were aware of this and may have had a surprise      
    of their own in the lines that they had prepared for Garry.       
   (OVER 2.8 MILLION moves in Fritz's opening library!!! And many of these lines 
    were possibly prepared with Garry Kasparov specifically in mind!) 

Bear in mind that Garry probably had spent MOST of his preparations in 
anticipation of Fritz playing 1.e4 on the opening move.  

     [ I was expecting the move:  1...Nf6{Diagram?}  
        with possibly a  "King's Indian Defense." 
        (Garry Kasparov is one of the world's leading experts 
         in this opening.) ]   


White immediately pressures the center and enters the main lines
of the venerable opening known as the Queen's Gambit.  

Of course any simple and/or developing move could be played by White 
at this point in the game. 2.c4 is not obligatory, just the most common move 
played at the Master Level. 

     [ I would be very interested to see how Garry would have handled a 
       more simple approach with a move like:  2.Nf3!?,   {Diagram?}    
       with possibly a Colle Opening. ]   


Yet another surprise! Garry plays the  "Queen's Gambit Accepted."     
I have a database completely dedicated to Kasparov's games. A search 
of that collection of games showed that - as far as I can determine - Garry 
has NEVER played the "Q.G.A." ...  from the Black side!!!
(I found dozens of examples with Garry playing the White side of this 
 particular opening!) 
 {Admittedly this database is a little old ... and contains ONLY serious or 
  slow chess. It does NOT contain fast games or blitz matches!}  

---> I put out a number of inquiries to some of my friends, and one has a very 
extensive database - which he maintains religiously. He said that Garry HAD 
played the Black side of "The Queen's Gambit Accepted," ... 
but only a few times. Apparently at least one (or two) of these were BLITZ games, 
(5-min.) and another may have been RAPID chess. (Game in 30 minutes.) 
I don't count these types of games, they do NOT go into my database of Garry's 
{collected} tournament (and also match) games. 
(This would explain why they were not in my database.) 

----> Apparently Garry had played the "Q.G.A." ... and one of the games was 
versus his main protagonist ... GM Vladimir Kramnik. (But to repeat this was a 
FIVE-MINUTE or BLITZ/Rapid game.) And he did win this game. {As Black.}  

   Now the question that begs itself is: "What is Garry up to?"    
   And, "Why is Garry playing this particular defense?"    


Without going into an extremely deep analysis of all the possibilities of the 
lines of the "Queen's Gambit Accepted," I will simply make the following 
statement. ----->
 I think Garry was heading or hoping for lines that feature: 

# 1.)  Wild sacrifices by White in the opening phase of the game;  
{There were a couple of lines popular about 10-15 years ago. One of these
 featured a number of sacrifices by White. It is entirely possible that Garry was   
 going to try and steer the game into one of these particular variations.}   

# 2.)  Very unclear or murky positions ... and positions that are not very 
thoroughly explored by theory;  

# 3.)  MOST Importantly! There are many lines in the Q.G.A. that feature an 
EARLY ENDGAME. Even the many commentators - like Hoffman - have 
specifically noted that the computers do NOT play endings really well. 
(The emphasis is on LONG-TERM planning, pattern recognition, ...  
  and NOT on tactics. And as Hoffman duly noted, game three showed 
  that the computer is deficient in these areas. GM Yasser Seirawan also 
  noted that some lines of the Q.G.A. feature early endgames as well.)  


Another added point is that the "opening book" of several versions of this 
program, (i.e. Fritz 7.0); featured bad or old analysis of this particular opening 
variation. I feel relatively certain that Garry  - and his team -  must have noticed 
this flaw and aimed to exploit it.



     [  The 'main line' ...  (of The Queen's Gambit - {Declined})  
        and the continuation that is certainly played the most  frequently is:  
        2...e63.Nc3 Nf64.Bg5 Nbd75.Nf3 c66.e3 Be77.Rc1 0-0;  
        8.Bd3, "+/="  {Diagram?}  
        and White maintains a small ...  but a rock-solid and time-tested edge.  

        One of the most recent examples I could find was the following game: 
        GM Giorgi Kacheishvili (2603) - T.(2) Petrosian (2259);  
        BCSA Open (Tournament) / Batumi, Georgia; (RUS) 2003.  
        (White won in 30 moves.);  


        By playing the moves:   2...c6!?3.Nf3 Nf64.e3 e65.Nc3 Nbd7;     
        6.Qc2,  "+/="  {Diagram?}   
        Garry could play the Black side of the same opening ... that was played 
        in Game # 1. (But with Garry playing White, not Black!)  

        So in effect, Garry would be playing himself!  

        Apparently Garry discounted this particular approach, although it was 
        enormously interesting to me and other chess fans around the world.  
        (But duly noting we are amateurs, and have little or nothing to lose!) ]     


This is pretty much the normal move for White in this position. 

It also is good because it: 
A.)  Controls the center;  
B.)  Develops a piece towards the middle of the board; 
C.)  It is also paving the way for a later K-side castling by White. 
       (As Fritz indeed will play in just a few moves.)  

     [ Another good - and very sharp move - is the following:  
       3.e4!?, {Diagram?}  with pressure on the center of the board. 
       This leads to a completely different type of game than was 
        played in this encounter. 

       I remember annotating several games with this particular line. 
       (One example.) (A more recent GM game.)  ]  


I watched the tape of this event very carefully. It is very clear that this 
is the exact move order of this game. (Many of the on-line sites have 
an incorrect move order.) 
Garry now gets a little tricky with his move order ... but to be honest 
it does not really matter (that much) what move order Black uses! 

     [ Black can also play: 3...Nf6; "=" (?)  {Diagram?}   
       {Which would eventually transpose back to book.}  ]    


The best move for White, although other moves are possible  
here for the first player.
(Besides the note below, White can also play a3!?, Nc3, g3, 
 and also e4 on move four.)  


     [ In the early days of this opening, the move: 4.Qa4+!?{Diagram?}  
        was thought to be the best reply. ]   


Again ... Garry is being very tricky with his move order here. 

In the old days - before hash tables - the computers would spend a lot 
of time, or even lose their way after little tricks like this. But here Fritz_X3D 
hardly pauses before cranking out the correct ("book") move in this position. 

     [  It is more usual for the second player to try the move:  4...Nf6{Diag?}    
        at this point in this line. {Again - transposing back to book.}  ]   


5.Bxc4 ,  
This is almost guaranteed to be the best move. It hits all four of the opening 
principles. It re-establishes material equality. And this piece will be posted 
aggressively on c4 - pointing somewhat menacingly in the direction of the 
Black King.  

     [ Playable is: </=  5.a4!?{Diagram?}        
       (Playable perhaps, but not great.) ]      


Garry plays with his pawns ...
perhaps hoping to throw the program off. But Fritz is not interested 
in even attempting to exploit Garry's move order - and bashes out 
its reply nearly instantly. 

     [ More normal is: 5...Nf6; first. (in this position) ]   


The program wisely castles.

This move is also good because the King is tucked safely away in the corner 
and a Rook is brought into play from out of the corner. 


     [ Also playable for the first player are the moves: 
        6.Nc3{Diagram?}   with a good game;  


       Or even the play:  6.Qe2!?{Diagram?}     
       which also gives a nice edge to the first player here. ]   


Garry finally develops this piece.  

It controls the center, gets a piece out, and prepares castling.  

It also heads back into the main line for the second player.  

     [ Garry probably considers the play of:  6...b5!?{Diagram?}      
        to be much too wild and too risky for Black in this position. ]    


7.Bb3,  {See the diagram ... just below.}       
The favorite move of Garry's (other) nemesis, GM Vladimir Kramnik. 



   The actual game position after White's 7th move. (kasp-vs-fritzx3d_g4_pos1.jpg, 96 KB)



Garry was OBVIOUSLY disappointed to see this move, I am relatively certain 
he was hoping for an entirely different kind of battle and/or middlegame position. 

At this point Garry put his head in his hands, one pundit on ICC said Garry looked, 
"as if he wants to cry." I do not doubt that Garry was both surprised & shocked - 
and probably more than a little disappointed at this turn of events. But I think that 
what was really happening was that Garry was mainly trying to recall his analysis 
 ... he may have had a choice of several key lines at this particular juncture of the 

Garry is also a good actor. He gave a small and quick, wry smile, (when this move 
was first played) - that GM Yasser Seirawan was quick to note. Yaz duly noted that 
this move - Bb3 - is a favorite ... and was pioneered by maybe the only player to 
defeat Garry in a set match. (GM Vladimir Kramnik) 

Seirawan goes on to note that Garry is probably thinking something like: 
"Did Kramnik help them set the opening's book up here, or what?" 

In the final say, Garry (now) mostly had the very difficult job of trying to recall every 
game he had ever seen in this line, and then make a mental decision of which 
variation was the most suited to his needs. 


     [ The most often played continuation is: 
        7.Qe2 b58.Bb3!?{Diagram?}  
        This is probably the strongest, and has supplanted the other move, (Bd3); 
        as the main line at this particular point. 

            ( Interesting was: 8.Bd3!?, which is the older line.      

               V. Chekhover - H. Kmoch;  Masters Tournament,      
               Leningrad, (U.S.S.R.); 1934.      

               (GM Artur Yusupov - who now spells his name Jussupow - played       
                this line as recently as 2002.) )        

        8...Bb79.Rd1 Nbd710.Nc3{Diagram?}  
        Once again, the opening is about center control and getting your pieces 

            (After 10.dxc5!? Bxc5, "~" {Diagram?} Black looks OK.)     

        10...Qb8; ('!') {Diagram?}    
        Here Black can play his Queen to nearly any reasonable square,  
        (c7, or b6). The second player also have several other moves, 
        including moving his KB.  

            ( Playable is: 10...Be7;  {Diagram?}  and Black should {eventually}     
              equalize ... with correct play. )      

        11.d5!? Nxd512.Nxd5 Bxd513.Bxd5 exd514.Rxd5 Be7{Diag?}      
        The end of the column.  

        15.e4 Qb7; 16.Bg5 Nb617.Rad1!? f618.Bf4 0-0!{Diagram?}   
         MCO says that this is better than taking on d5 ... although I imagine  
         Garry would love to have given this line a whirl!  

         ( KEY POSITION, # 1.)   (See the variation just below.)   

            ( Maybe Kasparov wanted to play the line:  
              18...Nxd5;  19.exd5 0-0;  20.d6,  "~"  {Diagram?}       
               when White supposedly has active play - but maybe Garry        
               feels he can keep the material ...  and try to win. )        

        19.R5d2 Rad8;  "="  {Diagram?}   
        Black has almost certainly managed to equalize here.  

        GM Larry Christiansen (2585) - GM Robert Huebner (2615);  
        (Bundesliga?)/Germany, 1992.  

        [ See MCO-14; page # 445, column # 1, and also note # (e.). ]  


        White can also play the move:  
        7.a4!?{Diagram}  in this position. The idea is obviously to prevent 
        Black from playing the freeing/space-gaining move, ...b5.  
        [See MCO-14; page # 447.]  ]  



7...cxd4; ('!')   
This is one of the more provocative strategies that Garry could adopt here. 
(White is given an isolated pawn - but also many open lines and also a lead 
 in development.)  

But almost no one - that I am aware of - has pointed out that this line avoids the most  
promising line for White  ...  simply by Garry using an intelligent order of moves.    

     [ Another favorite move here for the second player is: 7...Nc6!?;  "~"  {D?}  
        and Black  also gets a good game. 

        [ See MCO-14; page # 447, columns # 7 & 8, and note # (g.).]  ]   


Practically the only good move for Fritz_X3D ... in this position. 

     [ After the moves:  8.Nxd4!? e5!9.Nf3 Qxd110.Rxd1 Bg4;  "="  {Diag?}   
        Black has no problems in this position, and could even be a tad better. ]    


Once again - Black needs to start developing his pieces, 
although the try ...Be7 was certainly playable.  

     [ Was the move:  8...b5!?; {Diagram?}  playable in this position? ]  


The next few plays look to be the best ... for both parties here.  
9.Nc3 Be7;  10.Re1 0-0;  11.Bf4!, 
The computer played this move very quickly - almost instantly. It is very 
probably the best move, although a fairly new book I have on this very 
tricky opening ... advises White to try a3 in this position. 

     [ Interesting was the try:  11.a3!?, "+/="  {Diagram?}     
        when White also retains a small advantage from the opening. ]  


Garry too finds the sharpest ...  and probably the best move.  

     [ The continuation of:   </=  11...b5!?; ('?!')  12.d5! exd5;    
        13.Nxd5,  "+/="  {Diagram?}    
        yields White a very solid edge out of the opening phase. ]     


12.d5!?,  (Probably - '!')  
Absolutely one of the sharpest moves in this position ... and it could easily 
be the best. (Although this is not 100% certain.)   

The computer also whipped this move out very rapidly - it probably was 
all part of the prepared 'book' for the program. 

     [ Garry may have expected a line or continuation something like:  
       12.Bc2!? b5{Diagram?}    The TV commentator, GM Yasser 
       Seirawan commented on this particular position in one of his  
       "variations" during the game. He felt the computer would be quite  
       happy with this position. While this might be true, I think Garry came 
       well-prepared for this line as well.  

       13.d5 exd514.Qd3 g6!15.Nd4!?,  "~"  {Diagram?}     
        Here the official {book} verdict is that ... "White has very good play."  

       (From a book on this opening  ...  that is less than 3 years old.)  

       This must be someone's analysis. But no source is listed, and I could 
        find NO relevant matches in ANY database.  

       KEY POSITION, # 2.)    {After 15.Nd4.} 

       After days of research, I think this must have been one of the key 
       positions that Garry had been attempting to reach! ]     


12...Nxb3;  13.Qxb3,  {See the diagram just below.}       
The correct reply. White gets his Queen out and prepares to bring his 
other Rook to the center of the board.  

     [ </= 13.axb3? exd5; "=/+"  {Diagram?}  favors Black. ]   



   White just played Qb3. (kasp-vs-fritzx3d_g4_pos2.jpg, 93 KB)

(The actual game position after White plays 13.Qxb3.)


   (White has a lot of pressure here.)     


Garry now went in for a fairly long think, consuming over 20 minutes 
of precious clock time.  
13...exd5!;  {Diagram?}        
A surprise ... at least for the panel of commentators.  

Garry had been thinking for a fairly long time here. And during this time, 
GM Yasser Seirawan had stated, most emphatically and repeatedly, that the 
ONLY real move here for Garry was ...Nxd5 - preparing the same sacrifice  
that Garry had used against Kramnik.
  (And to be honest, several experts - like WGM Susan Polgar - also agreed      
    with Yaz on this matter.)     

Yet while Seirawan was so busy talking, Kasparov was deep in analysis. 
And Garry is not stupid. He has probably played many 'test' games against the 
computer. He knows that the computer is more than capable of finding the best 
move ... over and over again! It does not get tired and will not care how many 
times it has to search to find the best move. Therefore, Garry probably decided 
that giving the computer the Queen in this position was not the best choice.


     [ All  the commentators practically  GUARANTEED  that Black would  
       sacrifice his Queen in this position with the continuation of:  
       13...Nxd5!?{Diagram?}  This idea was originally an idea of the 
       ex-Latvian, GM A. Shirov.  (Who now lives in the country of Spain.)  

       14.Rad1 Nxf4!?15.Rxd8 Rxd8;  "~ {compensation}  {Diagram?}  
       and Black has good play for the Queen.  

       Kasparov  used this line to beat  Kramnik  in a "Botvinnik Memorial"  
       Tournament. (In 2001.) But I must emphasize that this was blitz, 
       (maybe rapid); and not slow chess. ]   


White's next move is a consequent play - and the best move for White. 
14.Rad1 Be6; ('!')  
While this is practically forced, for me it would have taken a lot of courage 
to play this move. For after ...Be6; Black's entire Queen-side is completely 

     [  Less attractive is:   14...Bc5!?15.Bg5, "+/="  {Diagram?}    
         and White is clearly better.  ]    


The best move - and apparently still a part of the program's prepared 
lines ... for it zapped this stoke out almost instantly as well.  

White obliterates Black's Queen-side, and only PERFECT play by 
Kasparov will prevent him from losing. 

This had all been played about a dozen times before - but not all of these 
contests were at the GM level of play. Probably the most significant game 
was:  GM V. Kramnik - GM V. Anand;  Melody Amber "Tourney of Stars"  
(rapid-competition, 10th round) Monaco, 2001.
{The game was a very hard-fought draw in nearly eighty moves.}  

     [ The machine could have also considered playing the move:  
        15.Nd4!?{Diagram?}   which also might lead to an opening 
        advantage for White. ]    


15...Bd6!; (TN?)  {Diagram?}        
Garry thought for a while ... and comes up with a large improvement 
over known 'book' theory.  

However, a friend with one of the largest private databases in the world  
told me that this is originally an idea of  GM Josif Dorfman,  and that he ... 
(he = Dorfman); has actually used it several times before.  

{Whether or not Garry was aware of this, I am not really sure. He seemed 
 to be working very hard at the board. But this could have been just for show.}  


     [ All the commentators agreed that the 'book' continuation was ...Bc5;  
        at this point. For example: 
        15...Bc5!?16.Be5 Qe717.Qxe7 Bxe718.Nd4 Rac8;   
        19.Bg3,  "+/="  (Maybe - '')   {Diagram?}   
        and White is certainly better.  

        GM Tomasz Markowski (2595)GM Jacek Gdanski (2543);   
        58th National Champ. Tournament / Warsaw, Poland; 2001. ]  



This pin looks to be the best move for White in this position.  

Up until this point, the computer had played almost all of its moves nearly 
instantaneously. (But I think this must have been the last move that was pre-
programmed into the computer's opening book. Because after this move, the 
computer begins slow down and "think.")

     [ 16.Bxd6!? Qxd6; 17.Qb3 Rab8; "<=>" ]    


16...Rb8;  ('!')   
The most active move Garry could play here ... and very nearly forced. 

Black has few viable alternatives, some moves - for Black, here - 
will lose horribly. 

And at long last, Garry seemed to have finally gotten Fritz_X3D out of its 
prepared lines - it settled down and had a nice 'think.'  

     [ After the (inadvisable) moves:  </=  16...Be7?!; ('?') 17.Nxd5! Nxd5   
       18.Bxe7 Rb8; 19.Qxa6 Qxe720.Rxd5, ''  {Diagram?}      
        White is clearly much better.  
        (Fritz probably wins here ...  without any undue difficulty.) ]    


Now the computer <thought> for close to five minutes. At this point Garry even 
asked the (FIDE) Arbiter if the machine's clock was even running! 
This move is practically forced, and is certainly the only move to yield White 
any advantage at all. 

GM's Ashley & Seirawan even wondered why the machine was taking so long 
to play this seemingly obvious move. The reason was is the computer has to sift 
through a number of alternatives, and examine dozens of lines ... even if they do 
not represent best play. (A human GM would probably play Qxa6 nearly instantly.)  

     [  </=  17.Bxf6!? Rxb7; 18.Bxd8 Rxd819.Re2 Bc5; "~"  

        </=  17.Qa7!? Qa5!; 18.Qd4 Bc5; "<=>"   ]     


17...Rxb2;    {See the diagram given, just below.}         
This must be the only good move here for Black. (And its not bad. 
Black grabs a Pawn and lands a Rook on the 7th rank.) 



   Black just played 17...Rxb2; - landing a  "pig on the 7th."  (kasp-vs-fritzx3d_g4_pos3.jpg, 91 KB)



     [ After the moves:  
         </=  17...Rb6?!18.Qe2 h6; 19.Be3, ''  {Diagram?}        
        White is clearly much better.  
        (And would probably win without too much trouble from here.) ]   


Play now proceeds down a fairly narrow and a relatively forced path ... 
for both parties.  
18.Bxf6!?,  (TN{Diagram?}      
This initially looks to be very promising - and was even the first choice of 
GM Yasser Seirawan. 
(Originally Seirawan thought Garry would have to re-capture with his 
  Knight-Pawn in this position.)  

But this move might take some of the pressure off Black ... and could allow 
him to equalize. But at the same time, Re2 had been played before ... and 
White was only able to draw. So it is an open question if - or how - White is 
able to actually increase his edge in this position.  

It is also entirely possible ... that this is the best move ... and that the computer, 
which can search over FOUR MILLION positions PER SECOND, (and thus can 
look nearly 10-15 moves deep in some positions); ... has found some resource 
for Black that has been otherwise over-looked by all the pundits!  

     [ Probably better was the following continuation, ...   
        but this is NOT 100 per cent certain:  
        >/=  18.Re2 Rxe219.Qxe2, "+/="  19...Be720.Qd3 Qa5;  "<=>"   {D?}     
        and yet Black had good play and drew shortly.  

        GM J. Dorfman - GM A. Kharlov National Championship Tournament / 
        Bordeaux, (FRA);  2001.  

        (An often overlooked fact is that Kharlov has worked for Garry's team in 
          the past ... and it is highly possible that Garry knew the "in's-and-out's"  
          of this whole line!) ]     


The best move ... and a surprise for the group of commentators.  
(The first time around, Yaz said if White captured on f6, then Black 
  would have to re-capture with the Pawn.)   

     [ Not nearly as good was the line: 
       </=  18...gxf6!?; 19.Qd3, "+/="  {Diagram?}    
       and White may be able to exploit Black's fractured 
       pawn structure. ]    


The best move for White.  

     [ Of course not the silly:  </= 19.Nxd5?? Bxd520.Rxd5??      
       20...Bxh2+; "-/+"  {Diagram?}    and White loses the Queen.  
       (As GM Y. Seirawan pointed out in his commentary.) ]     


19...Qxc3[];   {Box,}     
This was forced - Garry had to regain the piece in this position,  
or lose horribly.  

The correct move - White ditches the weak button on the a-file ...  
 and heads for high ground and the safety of a draw.  

     [ Possible was:  20.a4!? Ra2; "~" {Diag?}   with an unclear position. ]     


The next few moves are all pretty much forced in this position, now the team of  
commentators were beginning to see that the game could be drawn.  
20...Rxa2;  21.Nxe6 fxe6;  22.Qxe6+ Kh8;  {Box?}  {Diagram?}        
Practically the only good move.

     [ Simply horrible is:  </= 22...Rf7??23.Qxd5, "+/-"  {Diag?}   
        ... and now White is winning. ]    


23.Rf1,  ('!')   {See the diagram given here  ...  just below.}       
White - of course! - must protect the sensitive f2-square. But Rf1 was not the 
only move, I had seriously considered Re2 for White in this position. 

I think the computer's move - here, at this particular juncture - is best. 
The machine even thinks to keep all of its King-side Pawns ... and 
perhaps turn its attention to White's isolated d-Pawn in the near future.  



  White just played his Rook to f1 ... to defend the f2-square. (kasp-vs-fritzx3d_g4_pos4.jpg, 84 KB)

The actual position in the game ... just after White played 23.Rf1.




     [  Variation # 23W1.)     

        Also apparently playable was the move Re2, viz:  
        23.Re2!? Rxe224.Qxe2 d4!?25.Qe7 Qc826.h3{Diag?}    
        White must give some 'luft' to his King in this position.  

            ( Of course the computer would NEVER play:      
              26.Rxd4?? Qc1+;  and White will be mated. ("-/+") )     

        26...Rd827.Qe4 Qd728.Rd3 h6; "="  {Diagram?}   
        ... This position (also) must be drawn. (Similar to the game.);  


       Variation # 23W2.)   

        Very interesting was Qf7!?, but this is (hardly) no better than the 
        game, e.g.,   23.Qf7 Qc824.Qxd5 Raxf225.Re7 Qa6;  
         26.h3 Qb627.Kh2 h6"~"  {Diagram?}   
        but I cannot see either side winning this with best play.  


       Variation # 23W3.)     

        It seems that the capture, (with the Queen); on d5 ...   
        was good enough to draw as well.  For example:   
        23.Qxd5!? Raxf224.h3, "~"  {Diagram?}       
        and once again we have a situation similar to the actual game.  
        (A draw is obviously in sight.);  


       Variation # 23W4.)    

        It is also possible that taking with the Rook here would have 
        been acceptable as well:  
        23.Rxd5!? Raxf2{Diagram?}    
        The correct capture.  

            ( Not </= 23...Rfxf2???;  24.Rd8+, ("+/-")  {Diagram?}         
               and mate next move. )       

        One of the safer alternatives.  

            ( But not: </= 24.h3? Qg3; "/+"  {Diagram?}      
               and Black is clearly better. )        

        Now after the moves:  
        24...Qc225.Qg4 h6!?26.h3!?,  "~"  {Diagram?}   
        a draw is probably the correct outcome of this position. ]    



Now one might expect Black to play something like ...Ra5 and try to protect his 
d-Pawn. Garry comes up with something much better.  
23...Qc5!;  (Maybe - '!!')   
The best move ... with one play Garry actually accomplishes several 
different things:  
# 1.)  Puts the Black Queen on a better square. 
          (Nearer the center, and protects his Rook on f8.)  
# 2.)  The Queen now protects the Black Pawn on the d5-square.  
# 3.)  Garry sets up some really nasty threats to White's f2-square.  

     [ Possible was: 23...h6!? ]   


This move had already been completely discounted by GM Yasser Seirawan ... 
but it is definitely the best move for Fritz_X3D in this position. 
{White re-establishes material equality ... and the game winds down to a draw.}  

(Seirawan - at one point in his commentary - said this pawn ...  
  absolutely could NOT be captured! 

  He even referred to this button as: "A POISONED PAWN!")   


     [ White cannot play the other capture:   
        </=  24.Rxd5?!,  ('?')  24...Qxf2+!!{Diagram?}    
        Definitely the best move.   

          (Seirawan only looked at: </=  24...Raxf2!?; 25.Rxc5?? Rxf1#.)     

        25.Rxf2 Ra1+!;  ("-/+")  {Diagram?}   
        and White can give back his Queen and both Rooks ...  
        before he gets mated.   


        I actually expected the move:   24.Qe3!?, "~"  {Diagram?}      
        but what the computer played is much, much better. ]   



A super-shocker!    
(Black leaves his Queen hanging ... and grabs a ...  NOT-so-free Pawn.)
And it is almost funny.  

Many GM's were predicting that Garry would have to draw the ending of  
"Rook-plus-three-Pawns vs. Rook-plus-two-Pawns," but with Pawns only 
on just one side of the board.  

Garry finds something even better.
(Although to be honest, Yasser had predicted the capture on f2 here.)  

     [ The ending after the moves:  
        24...Qxd525.Rxd5 h626.g3, ''  {Diagram?}      
        is technically only a draw  ...  but Fritz (X3D) could 
        make Garry suffer for a really long time. ]   


I am sure this is best.   

     [  Variation # 25W1.)      

        Many commentators were saying that it was a mistake to play Qd8+  
        in this position. But that is simply incorrect, as the following analysis  
        will show. 

        25.Qd8+!? Rf8+{Diagram?}   
        This is forced.   

            (</= 25...Qf8??;  26.Qxf8+ Rxf8;  27.Rxf8#)     

        The next few moves are forced.  
        26.Kh1 Raf2;   {Diagram?}   
        This is virtually forced.  

            (Of course not: </=  26...Rxd8 ???;  27.Rxd8+ Qf8; 28.Rdxf8#)      

        27.Rxf2 Qxf228.h3 h629.Qe7, "="  {Diagram?}   
        and the position is completely balanced. (As in the game.). 


        Variation # 25W2.)     

       Naturally - White could NOT play:  
       </=  25.Qxc5?? Rxg2+;   {Diagram?}      
       White can only watch what happens now, as he is unable to protect 
       the key squares, and all of Black's moves come with check.  

       26.Kh1 Rxh2+{Diagram?}   
       Now White only has one legal move in this position.  

       27.Kg1 Rag2#{Diagram?}  and White has been mated.   

       Of course the computer would have NEVER fallen for this ...   
       unless there was a short-circuit or something!! ]   


The next few ply are best. (forced)
25...Qxf2+;  26.Kh1 h6;  
Black makes an 'escape square' ...  for his King, and guards against 
any possibility of a back-rank mate.  

At this point - it is obviously a draw. The computer is ignorant of this and 
plays on until the team of 'handlers' step in and recognize that Garry is 
correct - and the game is drawn.  

This is probably the simplest method. (And also best.)  

      [ After the move:  27.h3, "="  {Diagram?}    
        it is also time for smiles, a shaking of hands, 
        and the signing of the score-sheets. ]   


27...Kh7;  "="   {See the diagram just below {See the diagram just below the variation.}      
Draw agreed.  

      [ After the continuation of:  27...Kh728.Qd3+ Kg8!?{Diagram?}  
         A possible winning attempt?  

           ( The continuation of: 28...Kh8; 29.Qd8+ Kh7; 30.Qd3+,      
              30...Kh8; ("=")  {Diagram?}  is an obvious draw. )     

        29.Qd5+! Kh830.Qd8+ Kh7; "="  {Diagram?}  
        ... it is very obvious that neither side can make any progress. 
        After the same position has been repeated three times, either 
        party can claim the draw. ]     



   The final position ... of this very exciting contest.  (kasp-vs-fritzx3d_g4_pos5.jpg, 80 KB)

  (The final position after 27...Kh7.)  




An extremely exciting and very well played game by BOTH man and machine. 
Well done!! In my opinion, this is easily the best game of the entire match - both 
sides played without any mistakes.  

  This is easily one of the best draws I have seen in the last 50-75 years.  
   Both sides played practically perfectly!    (BRAVO!)  



  Copyright () A.J. Goldsby I  
  Copyright (c) A.J. Goldsby.  Copyright (c) A.J.G; 2003.  





   (HTML code initially)  Generated by the programChessBase 8.0   

   Chess diagrams generated by the program:  ChessCaptor 2.25  

Click   HERE   to go (or return) to the main/parent page for game number four. (# 4.) 

Click   HERE    to go or return to the  MAIN PAGE  for ... 
 the  Garry Kasparov-versus-Fritz_X3D match.  

Click    HERE     to go (or return) to my  main/home page ... for the ... 
eternal, fierce battle between HUMANS and THE MACHINES. (the boxes/computers)

   This page was first posted - in very ROUGH form! - just a few weeks after the completion of the match.  

Page (final) posting:  Friday; December, 19th, 2003.  This page was last updated on 06/07/12 .   


  Copyright (c) A.J. Goldsby I  

  Copyright () A.J. Goldsby,  2002 - 2005  &  2006.  All rights reserved.