There are probably many other complicated variations hidden in the position after 24 f4. I would like to draw the reader's attention to the importance of the open file for the Rook, be it either the f file or the d file >

So what does White gain by playing the move 23 h4?

To begin with, after 25 e5 White has a double potential threat of Rd7 and Rxf4 eg 25...Rxd1 26 Rxd1 threatening 27 Rd7 or 25...Rd6-d8 26 Rxd8 Rxd8 27 Rxf4.

Now suppose that White attempts to start his pawn breakthrough by playing 23 f4 and 24 e5 at once, without the preparatory move 23 h4.

The question will begin to be answered twelve moves later at move 27, but first the move 16 Nb1 frees the c pawn for the move 18 c3, and 19 Nb1-d2 redevelops the N to d2. Then White's Queen takes six moves to make her way to d3 via b1, c2, b3, c4 and e2.

Following the move 26 a3, the move 27 b4 begins to provide an answer to the question: the player who has the bishop pair is at greater liberty to advance his pawns. The pawns become a weapon instead of a target.

The advance gains White greater space on the Queen side and then the manoeuvre Nd2-c4-a5 begins a Queen side attack. White's Queen side attack ends up winning a pawn on the Queen side by means of the capture 34 Qxb5.

In return Black wins a pawn on the King side, but as before, Black's pawns are less free to advance than White's.

Petrosian relinquishes his bishop pair by 37 Bxe6 in return for crippling Black's King side pawns after the recapture 37...fxe6 when Black has doubled isolated pawns on e6 and e5.

After a few further moves Sax resigns, perhaps seeing no satisfactory answer to the advance of White's Queen side a pawn.

We can now, at the rish of repetition, answer the question of what White gains from the bishop pair. The first thing which White gains consists of increased liberty to move for his Queen side pawns. The advance of White's Queen side pawns leads to a Queen side attack which wins a pawn on the Queen side. Black is able to win in return for a pawn on the King side.

Then however White is able to use one of a second power of the bishop pair to cripple Black's King side pawn majority, by relinquishing a bishop for Black's N.

Consider first what it gains. It inflicts an isolated c5 pawn on Black and the capture 17...Nxe4 concedes a tempo to White's Queen.

Now consider what this capture concedes to Black. 17 dxc5 concedes a tempo as well as opening the b file for Black's Rook. In addition Black gains by far the more powerful minor piece and the White Queen will have to move again after ...Bb7, and that makes a second tempo lost.

The conclusion is that on balance the capture 17 dxc5? concedes almost the value of a pawn to Black.

Topalov's play which follows this offers a lesson in what happens when a player wants more than a draw from an equal position and attempts to start unsound attacks from it, attacks which are not based upon positional advantage.

21 f3? disturbs the King side pawns without necessity. On 21 Nc4 Qg5 22 e4 makes way for Rg3. However that may allow Black to equalize and possibly for this reason Topalov avoids it.

23 g3? disturbs again the King side pawns without necessity. 23 Nf1 may be perfectly safe, but Topalov does not want perfect safety. He wants to play for a win.

After 25...Ba6 it is probably Black who stands better and White who should think of how to equalize, perhaps by 26 Nc4. Topalov does not want it. He plays to win by 26 Ra3. This is an absurd attempt to win from what is fact a disadvantageous position. It is getting serious now because he is playing to attack when he has no advantage. This could easily increase Black's advantage first to a strong degree and then to a winning degree.

27 Nb3? may be the last mistake, persisting in attempting to win from a disadvantageous position. This may increase Black's advantage from a substantial one to a winning one. Topalov has finally gone too far attempting to extract a win from an equal position.

After 27 Nb3 is the move 27...Rc7 necessary?

Suppose that instead of defending the c5 pawn by 27...Rc7 Black offers White the c5 pawn by 27...e5.

This move begins an attack based upon positional advantage. Providing that the attack is not disproportionate to Black's degree of advantage, such an attack is sound and will probably succeed against any defence. In that case it will be therefore too then to save White's game.

On 27...e5 28 Nxc5 Rc7 29 b4 e4 White is in some trouble.

On 28 Qxc5 Qe6 29 e4 g5 30 g4 Bxe4! 31 fxe4 Qxg4+ 32 Kf1 Rd1 is mate.

The advance 30..f5! also begins an attack based upon positional advantage, but as White has been given time to play the moves 29 Nc4 and 30 e4, Black's advantage has become reduced. The attack is still sound providing it is not disproportionate, and in that case it will probably succeed against any defence. However it may now take Black greater trouble to win, if he can do it at all.

This suggests that during games one and two at the very least, and to some extent throughout the match Anand was, to some extent, handicapped and not playing at full strength, and that this was caused by the Bulgarians' refusal of a three day postponement.

This suggests that if Anand had suffered no such handicap, Anand would have won his Sofia 2010 match with a greater margin than the margin with which he won against Kramnik, for example by the score of 6.5 to 3.5, and that it would not have been surprising if Anand had won the match by the score of 6.5 to 2.5.

Anand had a chance to gain the advantage and possibly win in every one of the games where he had White, and drawing chances in every game where he had black, including game one.

If the Bulgarians had not opposed Anand's application for a three day postponement and Anand had been granted the three day postponement, we do not know what difference a three day postponement would have made to Anand's choices.

Thus if Anand had been granted the postponement it is not beyond the bounds of possibility - and therefore not ridiculous- that after game eight the score would have been 6 to 2 in Anand's favour and that Anand would have won the match by the score of 7 to 2 after game nine.

Remember that Anand has said that he mixed up his move order in games one and two, and it seems reasonable to assume that this was because he was not granted the postponement. Who knows what the effects of the Bulgarian opposition to the postponement were in the other games?

Now consider the following question: If Anand plays a match with Kasparov now, what will the result be?

In 1995 Kramnik said that Kasparov was not really that much better than Anand, if he was better at all, but that Kasparov was better prepared.

Even so, the match was not as one sided as all that. Anand in fact took the lead in game nine, and after Kasparov won game ten, the score was even.

Then however Anand scored just half a point out of the next four games. Even here, Anand had not stood that badly in the games which he lost, but after the loss in game ten Anand blundered in game 11, and may have become demoralized for at least a few games.

However Anand recovered sufficiently after game 14 to draw the remaining games.

The Anand of 2010 will be very different from the Anand of 1995. At the very least, Anand is likely to prepare as thoroughly as Kasparov is accustomed to prepare.

Following the sequel 24...Qd6 25 Ne5 Re8 26 Re3 Rd8, the move 27 Nc4 begins to provide an answer to the question. Black cannot win a pawn on d5 because with his back rank exposed to the potential threat of mate, his Rook on the back rank is overworked.

This suggests the question of why.

One possible answer is that the choices which White has made before Black's seventeenth move have provided well for the consequences of the move 17....Ne5 but they have exposed White to the consequences of the move 17...Nf8.

If White waits, Black can play 16...c5 keeping White's N out of d4. This suggests playing 16 Nd4 first. Then on 16...c5 17 Nf5 the N has made use of the central square d4, using it as a springboard to the outpost f5.

One way for Black to gain the worse of it is 17...Re5 18 Qd3 Bc8 19 Ne3 Be6 20 Rfd1 Bh6 21 f4 Rh5 22 Bf3 Rh3 23 Nxd5.

Instead of this Botvinnik offers a second pawn by 16 e4?! and it is not obvious just exactly what Botvinnik gets in return for it. Perhaps Botvinnik expects mistakenly to create some eventual threats which will regain the lost material at the least, threats which Bronstein turne out however able to answer.

The move 18..Qf8!! is the first of a pair of moves with the Queen, of which the second 30..Qd6 will centralize the Queen, transferring her to the central d file a dozen moves later. However the Queen is not the only Black piece to get centralized in this way.

The remainder of the game looks like a triumph of centralization. Bronstein centralizes the whole of his officer corps, moving each and every one of his pieces to a square in the central files.

After 29...Bd5 Black threatens 30...Bxf3+ when the N on d4 will be pinned. 30 Re1 is one way to unpin the N. On 30 Rf1 Bronstein in his book "The sorceror's apprentice" quotes a variation given by Keres: 30 Rf1 Qd6 31 Nxf5 Bxf3+ (not 31...Qxc7?? 32 Qf6+ Kg8 33 Qg7 mate) 32 Rxf3 Qd1+ 33 Kg2 Rd2+ 34 Kh3 Rxh2+! 35 Kxh2 Qg1+ 36 Kh3 Qh1+ 37 Kg4 Rg8+ 38 Kh5 Rg5+ 39 Kh6 Rxg3+ !! displacing the R on f3 from its obstruction of the d1-h5 diagonal so as to be able to check with the Black Queen on the diagonal d1-h5 and so conclude the King hunt: 40 Rxe3 Rg6+ 41 Kh5 Qd1+ 42 Rf3 Qxf3+ 43 Qg4 Qxg4 mate

<sevenseaman: Where do such brilliant ideas come from?>

The question may refer to either to the final combination or to the preparations leading up to it, depending on whether by the brilliant ideas is meant either the final combination at the end, or the preparations leading up to it

In the former case the sacrifice 30 Qh8+! draws the Black King away from his defence of the Rook on f7, and at the same time draws the Black King on to a square - the square h8- which can be forked.

The N on d6 is able to deliver a fork from f7, one of whose targets is the square h8. When the N on d6 takes the Rook on f7, it delivers a fork to the Black King on h8 and the Black Queen on g5.

In the latter case more than one thing can be said.

One thing is that Petrosian makes two exchange sacrifices both of which advance further the development of his pieces. The move 21 Ne3 makes way for the Rook on a1 to come to f1, and the sacrifice 34 Rxf4! removes the N on f4 which hinders the check Be6.

A second thing is that the three principal White actors in the final combination, White's Queen on b2, the B on e6 and the N on d6 have all reached their posts by development followed by further development.

The Q has developed first to c2 and then to b2, the bishop has developed first to e2, then further from e2 to g4 and then developed further to e6, and the N has developed first to d2 and then developed further from d2 to e4 and then developed further still to the point d6.

A third thing is that the long range pieces such as White's Queen and Bishop have developed by occupying open lines whilst the shorter range Knight has developed to a central point and then advanced further.

The Queen has reached the long diagonal while the King's Bishop has occupied first the h3-c8 diagonal and then the a2-g8 diagonal.

A fourth thing which can be said is that every one of these three White pieces has developed finally to a square whence it attacks points in the enemy's camp, namely, the neighbourhood of the Black King.

The bishop on e6 attacks the points f7 and g8, the N on d6 attacks the point f7, while the Queen on the long diagonal attacks the points g7 and h8.

Lasker does not specify exactly how Black will gain ground and exactly which ground Black will gain or try to take, and one game which Lasker lost from a similar pawn formation is the game Capablanca vs Lasker, 1921

One conceivable way for Black to gain ground is to play ...Qf6 and then ...Nf4 if White advances his e3 pawn to e4. White has not played e4 yet, however, and this suggests that Lasker believes that the exact ground which Black either gains or is advised well to try to take depends on White's choices.

Furthermore Capablanca says repeatedly in his book "Chess fundamentals" that chess cannot be learnt from a book alone, and if Lasker holds the same view, this suggests that Lasker expects his readers to gain practice in playing from the Black side of this position, and trying to gain ground.

One idea is to make a decision about which Rook to play and then to choose the other one, because one always chooses the wrong Rook. Admittedly this cannot be called a true answer.

Then Black cannot run with the c6 pawn after offering the c5 pawn and White does bring his King into play in time.

Yet Black seems to play a king ahead for at least some purposes, because what is White's King doing on the Queen side?

One answer is that Black's King requires at least a few moves to win the c3 pawn and White's King gets into play in that time.

Another more interesting answer is that Black's Queen's bishop is overworked. If it has to defend the f7 pawn, it cannot allow itself to get captured on a2 as well.

So we have this interesting paradox: White's King, by attacking the f7 pawn on the King side, defends the Queen side because he prevents Black's Queen's Bishop from taking the a2 pawn and inviting the sacrifice Rxa2! removing the piece defending the f7 pawn which keeps back the e6 pawn.

White can play first the in-between move (or zwischenzug or intermezzo as this type of move has been termed) 19 Rad1! attacking Black's Queen.

The Black Queen is then short of squares as she lacks an alternative square from which to defend both the c7 pawn and the N on c6 as well.

On 20...Qf7 21 Qxc3 it turns out that not only is the B on c3 a target for White's Queen on the c file, the N on c6 is as well.

On 20...Qe6 21 Bxc7 the White bishop both retakes a pawn and escapes.

However the move 16...h6? disturbs the King side pawns without necessity so that the h pawn joins no longer its colleague on f7 in the task of defending the square g6, and the f7 pawn is left with the task of guarding by itself the square g6.

It is not obvious why the f7 pawn will be be pinned by White's queen's rook on the seventh rank, a rook which attacks indirectly the g7 pawn. The reason is that White's Rook will not act on its own but in concert with two other White men which will join the Rook's attack on g7 in two different ways.

White's pawn on f6 and a N on f4 will join this attack and act in concert with the Rook as follows.

The pawn on f6 will join the attack on g7 after the f7 pawn which obstructs White's queen's rook on seventh rank defends the point g6 has been diaplaced to g6.

The N on f4 will invade the point g6 after the f7 pawn which defends the point g6 has been displaced, and a Rook arriving on the point g7 has taken away the flight square g8 from Black's King.

The White Q on g6 serves to displace Black's pawn on f7 so that after 20...fxg6 21 Rxg7+ Kf8 22 Nxg6 mate becomes possible.

The lesser reason for this is that the first move of the pair, the move 15...Nxd4?! opens the c file. The greater reason is that the second move of the pair, the move 16...Qxd4, exchanges the Queens and so removes them from the board.

The first result of the removal of the queens is that White is rid of his most serious liability, an uncastled King which is exposed to attack.

The second result is that after the exchange of Queens, the evaluations of the position become all reversed, as if by magic.

All of White's liabilities become transformed into assets, and all of Black's assets become transformed into liabilities. Targets for Black targets become transformed into weapons in the hands of White.

The c file, instead of exposing White's King to attack with queens on the board, becomes after the exchange of queens, a weapon in White's hands. White's advanced King side pawns become transformed, after the exchange of queens, from targets into a weapon. White's King becomes transformed, after the exchange of queens, from a target into a weapon.

With the c3 pawn transferred to d4 and the White King safe, Black, instead of having prospects for a Queen side attack or for a King side attack, has passed the initiative on both wings to White, who threatens now to use both the c file and a King side pawn advance to attack Black.

Instead of 15...Nxd4, The right move is 15..0-0-0 with ...f6 to follow after which Black seems to gain the advantage. White seems unable to maintain even equality against Black's potential threats of ...f6 starting an attack along the f file and an attack upon the d4 pawn by the manoeuvre ...Ng6-e7-c6.

Possibly one way of putting this is to say that the effect of the pair of moves 15...Nxd4?! and more so that of 16...Qxd2+? is to remove White's most serious liability, an unsafe King. The removal of this liability makes the difference between Black having the advantage and White having a possibly winning positional advantage.

However they may have been mistaken, and the resident commentators ( IM Lawrence Trent, and either GM Stephen Gordon or GM Chris Ward) may have been right to believe that White stood considerably the worse because his N was placed very badly.

Here is one sequence. On 31 Ng1 Nc5! Black's N is in play, tying White's Q to the defence of the d3 pawn, whilst White's N on g1 does nothing. On 32 Qe2 Rg8 threatens 33...Nxd3. On 33 Nf3 h3! 34 gxh3 Ne4!! puts the d4 and f5 pawns to work, preparing to replace the N on e4 with the f5 pawn, forming together with the d4 pawn a central pawn phalanx. On 35 dxe4 fxe4 36 Nh2 d3! 37 Qf1 e3+ 38 Nf3 exd2 looks like a win.

To touch on just one aspect of the game, it is not obvious how the move 15...a5 will make the a5 pawn a target for White's Rook even after White plays 18 Nc4, 20 Nb6 and induces 20...Rb8 so that Black's R is unable to defend the a5 pawn.

The analysis suggests that the winning method is for White to set up a zugzwang with the White g pawn on a black square, and whether the White g pawn is on g3 or on g5, with the following type of formation for the pieces: White's K is in front of White's g pawn ( on g6, if the pawn is on g5, or on g4 if the pawn is on g3) and behind the remaining Black pawn, Black's King is on a rank behind the g pawn on a black square, and so a knight's move away from White's King, on the f or h file, and White's B is placed so that it controls the White squares around the Black King, e6 if the black King is on h4 and d5 if the black king is on f2. Then with Black to move, Black's King has to abandon his attack on White's g pawn.

With that explanation, here is Karsten Mueller's analysis: <69.Kg3? Kramnik allows the king to penetrate too deeply. [Hiarcs 13, running on a 12-core machine and providing added information to the live commentary, calmly spat out a mate score with an irreductible winning line. Here it is with some added analysis to clarify. 69.g5! hxg5 (69...h5 70.Kf4 Kd4 71.g3 Kc5 72.Kf5 Kd6 73.Ba2 Ke7 74.Kg6 Kf8 75.Kh7 ) 70.g3 opening the gates to the deadly endgame weapon: the Zugzwang, which takes the day once more. 70...Kd4 71.Kg4 Ke3 72.Kxg5 and here all attempts to defeat the triangulation with the bishop will fail. 72...Kf3 (72...Kf2 73.Kf4 Kg2 74.g4 Kh3 75.g5+ Kh4 76.Kf5 Kh3 (76...Kh5 77.Bf7+ Kh4 78.Ba2 Kh5 79.Be6 Kh4 80.Kg6 ) 77.Bb3 Kh4 78.Ba2 Kh5 79.Be6 Kh4 80.Kg6 ) 73.Kh4 g6 74.Bf7 g5+ 75.Kh3 Kf2 76.Kg4 Kg2 and now a triangulation with the bishop wins it. 77.Ba2 Kf2 78.Bd5 ] >

One more thing. Ilias Kourkounakis says in his chessbase article that <Even if the K+B+Ps vs K+Ps endgame is winning, it is clear that the present balance of power wins more easily (for example, by first eliminating the enemy a-Pawn). On the other hand, while the present balance of power is surely winning, it is not 100% sure that the same is the case with the K+B+Ps vs K+Ps endgame. Therefore, not liquidating would maximize his chances, while liquidating might decrease them.>

1. The move 25 Bd3 uncovers his heavy pieces on the e file and induces the manoeuvre ...Bf6-e5.

2. The exchange 27 Bxe5 draws the d6 pawn on to e5.

3. The advance 32 f4 attacks the e5 pawn and induces the advance 32...f6.

4. Upon the advance 33 g4 Black can't remove the g pawn as the h pawn is pinned.

5. With the advance 35 g5 White's f4 pawn and g5 pawn are attacking Black's e5 pawn and f6 pawn respectively.

6. The exchange 37 gxf6 exchanges the g5 pawn for the e7 pawn, removing one obstruction from the e file

7. The exchange 38 fxe5 removes the f6 pawn defending the pawn on e5.

8. The move 39 Nd3 overpowers the last obstruction on the e file towards Black's King.

This compensation consists of prospects of attack arising out of White's greater space and lead in development.

The group of moves 7..Nc6, 8...Nb4 and 9...Qd5 induces White to exchange Queens and so to relinquish the prospects of attack which make up White's compensation for the Bishop pair.

Perhaps the move 7 Qd3 is not the best, as it gives Black's QN a tempo for the attack ...Nb4 following ...Nc6.

The reason is that the move ...Nc5 imposes a second task upon the pawn on d6, that of covering a N on c5.

If in the position after 11...f5 we take the black knight which is on b6 and place it on the square c5 instead, the d6 pawn is overworked as it can't defend both a N on c5 and the e5 pawn as well.

White can then win a pawn by 12 exf5 Bxf5 13 Bxc5 dxc5 14 Nxe5.

With the move 9...Nb6 the knight may appear to "do nothing" when compared a N on c5 although even this is not quite true: it does free Black's QB, and with tempo.