3... g6

click for larger view

"A double fianchetto defence is too much of a luxury...White [now] profits from his advance in development to gain control of the centre. Instead, Black has at his disposal the simple and good 3...c5..." (Golombek)

"Simpler was 3...c5, but Keres doesn't like symmetrical positions."

"With this move, Black heads for the King's Indian Defence, where his extra tempo b6 does not have any significant importance. Therefore simpler was 3...c5...Black would achieve strong pressure against White's centre." (Keres)

"The best results reaction was natural play in the centre--3...c5...But the leader [Keres] was aiming for more complicated play, and for him a King's Indian set-up (also with an extra tempo!) seemed most appropriate." (Kasparov).

4. e4

click for larger view

"Threatening e5, hence Black's next move." (Golombek)

4... d6

click for larger view

"If 4...Bg7 5. e5 Ng4 6. d4 followed by h3 could become awkward." (Keres)

"...[picking up on Keres' line and reaching a different conclusion] 6...d6 7. h3 Nh6 8. Nc3 [8. Bd3 looks better, while Golombek's 8. Bf4 allows Black to equalize with ease--KEG] dxe5 9. Nxe5 Bb7 [9...Nf5 is even better--KEG] Black's problems are quite resolvable...." (Kasparov)

Kasparov also gives 4...c5 ("!?"--Kasparov) as "beginning an immediate fight for the centre" which he says is a "quite acceptable way of developing." The question in this line is the status of the Black K-side Knight after 5. e5 Nh5 which seems OK after 6. d4 cxd4 7. Qxd4 Nc6 of after 6. Be2 followed by 8. Nc3.

5. Nc3 Bg7

click for larger view

Now came the remarkable follow-up by Reshevsky:

6. d4!

click for larger view

Can White afford this seeming loss of a tempo:

"Reshevsky shows a bold disregard for the elementary principle of opening strategy, that one should never waste Pawn moves, [but does so] in this case with good reason. For we have now arrived at a King's Indian Defence...with the different that Black has in addition played b6. But the Pawn move is quite out of place in the King's INdian, and merely means that Black has weakened his Q side." (Golombek)

"It looks as if White has simply lost a tempo for this is the King's Indian Defense with b6 as an extra move for Black. Right--but the extra move means an extra weakness." (Kmoch)

"Now a King's Indian has arisen in which Black has gained the move b7-b6. But it is a gain of doubtful value." (Euwe)

"This advance, though a loss of tempo, is sooner of later forced if White wants to take up the battle for the centre. The game has now clearly transposed into the King's Indian Defence with the extra move b7-b6 for Black, which...does not offer him any major benefits." (Keres)

"This move is necessary, but now play...switches to King's Indian lines, and the tempo spent by Black on b7-b6 does not play any role." (Kasparov)

Also important was the psychological strategy of White:

The machinations described in my earlier posts and the potential weakness created by b6 notwithstanding, Keres' position to this point was at worst only slightly inferior. The question now was how Keres would handle--and how Reshevsky might try to attack--the Black double fianchetto arrangement.

6... 0-0
7. Be2

7. Bd3 looks more to the point, but Reshevsky had decided to build on his small edge slowly and to avoid any theoretical lines that might play into Keres' mental data-base.

The position was now:

click for larger view

7... Bb7

I was surprised to read the criticism of this entirely normal move. Some of the commentators, however, remained so skeptical of the double fianchetto set-up that they recommended dispensing with this seemingly obvious placement of Black's light-square Bishop.

Thus, Kmoch preferred 7...Nbd7 followed by e5 and deferring a decision on how to develop the Bishop.

Kasparov in this commentary on this game was more emphatic, saying that 7...Bb7 was "not in accordance with the logical development of the black pieces" and recommending instead 7...c5 and if 8. d5 then 8...e6. Kasparov ultimately sought to exploit Black's extra tempo "in a Modern Benoni set-up, where after an exchange of pawns on d5 the [Black] bishop could come out to a6."

This approach seems debatable at best. While the White pawns are likely to block the Black Bishop on b7, the pressure on the White center surely counts for something. And as for developing the Bishop on a6, that square might be a useful route for the b8 Black Knight in Kasparov's variation: 7...c5 8. d5 e6 9. 0-0 exd5 10. exd5 Re8 11. Re1 after which most natural for Black seems to be either 11...Bf5 or 11...Na6 followed by 12...Bf5.

I am certainly not saying that Kasparov's idea is without merit. But the criticism of Keres' move--or the suggestion that it had much to do with his loss of this game--seems a bit over the top.

After 7...Bb7, the position was:

click for larger view

8. Qc2

Kasparov praises Reshevsky's patience in declining to play the superficially indicated 8. d5. Indeed, the strengthening of the White center with the text in lieu of 8. d5 e6 9. 0-0 exd5 10. exd5 c6 11. Bg5 cxd5 12. cxd5 (to give just one plausible line after 8. d5) does look like a better approach to the unique opening position in this game.

8... e5

"Threatening 9... exd4 10. Nxd4 Nxe4."

8...Nc6 and 8...c5 were reasonable alternatives. After the text, the position was:

click for larger view

9. dxe5

Should Reshevsky instead have played 9. d5? The commentators are divided on this issue.

In favor of the text were Kmoch and Keres:

"Very well played." (Kmoch)

"Black's extra tempo in developing the Bishop to b7 at last has the effect that White, in contrast to the normal King's Indian Defence, immediately has to resolve the tension in the center." (Keres)

Keres went on to concede that even with the text, White gets no real edge, but argued that 9. d5 Nbd7 followed by Nh5 "would have given rise to a position where Black seems to have the more favorable prospects."

Kasparov, Euwe, and Golombek disagree, and prefer 9. d5.

After 9. d5, Golombek, like Keres, only considers 9...Nbd7. But the crucial line after 9. d5 is the one given by Kasparov: 9...Nh5 10. g3. Kasparov notes that the resulting position was akin to those handled so superbly by Petrosian, and that White is clearly better.

In fact, White has some edge after both 9. d5 and after the text, and the choice is primarily one of style. To the extent there was anything wrong with Reshevsky;s 9. dxe5, it was his questionable continuation on move 10.

9... dxe5

click for larger view

10. Be3

As Keres, Euwe, and Golombek noted, 10. Nxe5? would have been a serious mistake, since after 10...Nxe4 "the advantage would tilt to Black." (Keres). A likely continuation might be 11. NxN [not Golombek's 11. Nxf7? which loses a piece to 11...BxN+] BxN 12. 0-0 Nc6 or maybe 11. Nf3 NxN 12. bxN Qf3 13. Bb2 Re8; with Black holding the edge in either case.

The text gave Keres a chance to turn the tables a bit. Reshevsky should have played 10. Bg5!, a move mysteriously unmentioned by any of the commentators.

After 10. Be3, the position was:

click for larger view

10... Re8

Here Keres missed a chance to solve his opening problems. As Keres himself later said:

"From now on, Black starts to play inexactly and finally finds himself in insuperable difficulties."

What should Keres have played?

A) 10...Ng4?, as Golombek noted, "would only help White, e.g., 11. Bg5 f6 12. Rd1 Qe7 13. Bc1 [even better and leading to a serious edge for White is 13. Bd2--KEG]."

B) 10...Nbd7 was the choice of both Euwe and Golombek. But they only considered the variation 11. 0-0 (11. Rd1 is far stronger and gives White an edge even beyond what he got in the actual game) c6 (11...Qe7 is better) 12. Rfd1 Qc7 (inferior to 12...Qe7). Keres, surprisingly, repeats this same flawed line in his commentary. In light of 11. Rd1, this alternative cannot be correct. (In fairness to Keres, he did not give 10...Nbd7 as his first choice).

C) 10...Qe8 is superior to choices A and B, but is not as strong as:

D) 10...c5 looks best by far, and has been championed by both Keres and Kasparov. Kasparov said that 10...c5 "would have given [Black] a perfectly good game." In fact, 10...c5 was perhaps even better than Keres or Kasparov believed. Keres gives the continuation 10...c5 11. Rd1 Qc8 [11...Qe8 is better] 12. Nd5 Nc6. Kasparov gives a further continuation of 13. Bg5 NxN (13...Nh5 is best) 14. cxN Nd4 15. NxN exN 16. 0-0 f5.

11. 0-0

11. Rd1 was strongest.

After 11. 0-0, the position was:

click for larger view

11... Nbd7

Keres could still have obtained nearly equal chances with 11...c5 (rather than the wimpy text-move), as pointed out by both Keres himself and later by Kasparov.

Golombek trashed 11...c5 as an alternative, but his analysis was flawed:

"The ambitious attempt to establish a Knight on d4 by c5 and Nc6-d4 fails against 12. Rfd1 (Kasparov's 12. Rad1 is better, but also yields only a modest edge for White--KEG) Qe7? [weak, Black is fine with 12...Qc8--KEG] 13. Bg5 and White gets his Knight to d5 first [and indeed, in this line, could just play 13. Nd5 immediately--KEG]."

So why didn't Keres play 11...c5. Fortunately, we need not guess:

"...Black could have proceeded with 11...c5 followed by Nb8-c6-d4 if he wished, with a roughly equal game. With the text-move, Black is aiming for more, and hopes to utilize the weakness of the d4-point with the manoeuvre Nd7-f8-e6. But when taking into consideration that White is somewhat ahead in development, and that Black himself also has some weaknesses on the queenside, he cannot have any great hope of success of his plan."

12. Rfd1

click for larger view

12... c6

Everyone sees this as a weakening move. <RookFile> suggests that the need to play c6 shows the inaccuracy of the "committal" 7...Bb7.

Kasparov said that: "Black can no longer get by without this move."

Golombel bemoaned the resulting weakness caused by 12...c6, but said it was necessary in order to guard against the threatened Nd5.

But was 12...c6 necessary? In fact, Black does better to play 12...Qc8, after which he would only be marginally worse and would have reasonable counter-play after 13. Nd5 Nh5 or 13. h3 Nc5.

After the weakening 12...c6, Black's position was beginning to look suspect:

click for larger view

<....Reshevsky immediately diverges from the theoretical oaths....>

As I should have mentioned in my last post, the quotation concerning why Keres did not play 11...c5 was from Keres himself. That is how we know what Keres was thinking.

After 12...c6, the players were already heading for time trouble, having used nearly 40% of their time to reach move 40 after only 12 moves had been played:

Reshevsky: 0:56
Keres: 0:52

I should note that all precise time references I provide come from Kmoch's commentary on the game.

13. b4

"!"--(Keres)(Kasparov)(Kmoch)

"White has completed his development and with the text-move initiates an attack against Black's Queen-side. This constitutes the only positionally reasonable continuation here." (Keres)

In fact, although not mentioned by any of the commentators, 13. c5 was a more immediate and even stronger means of Queen-side attack than the text, especially in light of the pin on the d-file. A likely continuation is 13...Qe7 14. cxb6 axb6 15. a4 with even stronger pressure than with the preparatory 13. b4.

After 13. b4, the position was:

click for larger view

13... Qe7

"?"--(Keres)(Golombek)

The commentators except Kasparov concluded that Keres should have played 13...Qc7. But is this all that clear? While I agree that 13...Qc7 was probably marginally better than the text, it seems clear that: (i) Keres' position was not all that bad after the text (and he had plenty of later chances); and (ii) Keres would not have been that much better off with 13...Qc7. Moreover, (iii) 13...Ng4 was probably best for Black here.

Turning first to the case against the text:

"This loses a tempo as the maneuver Nf8-e6 is indicated whereby Black must care for his e-pawn." (Golombek)

Keres made a similar comment in his analysis.

But as will be seen, even after 13...Qe7, Keres could still have played 14...Nf8 and then 15...Ne6 (as I discovered to my initial surprise when feeding the position to Fritz and Stockfish).

As for the merits of 13...Qc7, all of the commentators assume that White would reply either 14. b5 or 14. c5.

Keres said after the game that he did not play 13...Qc7 because he feared 14. b5. he later concluded, however, that Black would be fine with 14...Nf8 15. bxc6 (somewhat better is 15. Rac1--KEG) Qxc6. Kasparov later contended that White would obtain a significant edge here with 16. Nd5, but Black is basically OK with 16...N8d7 and after Kasparov's 17. a4 ["!"] 17...Ng4. But more fundamentally, after 14. b5, Black can simply play 14...Ng4 or even 14...Bf8 (Kasparov's suggestion).

The real problem with 13...Qc7 is that White has better choices than 14. b5, including:

(A) 14. c5 and then after Keres' suggested 14...Ng4 White has the better game with 15. Bd2. and if instead 14...bxc5 (probably best for Black) 15. bxc5 Ng4 16. Bc1. And if 14...b5? White has the pleasant choice between 15. Bxb5!! (15...cxB 16. Nxb5 c6 17. Nd6 Re6 18. Ng5) and the simpler 15. a4.

(B) Slowly building his position and preparing for a later attack with 14. Rab1 or 14. h3.

Lost in all of the commentary is the value of 13...Ng4 which puts the question to White's Bishop after which Black can perhaps play 14...Qc7.

After Keres' actual 13...Qe7, the position was:

click for larger view

14. Rab1

As Keres and Golombek both noted, 14. b5 would have been premature in light of 14...cxb5 15. cxb5 Rac8, after which Black is clearly better.

14... Nf8

Consistent with the plan noted above of Nd7-f8-e6. But was this really best? The critical lines for Black here are those stemming from 14...Ng4 [15. Bg5 Ngf6 16. h3 (to prevent a future Ng4 by Black) Red8 17. Be3]; 14...Red8 [15. a3 Nh5 16. g3 (or 16. c5)]; and 14...a5 [15. bxa5 Rxa5 16. h3]. Since White retains an advantage in each case--as he does with the text--all of these plans seem reasonable.

After Keres' 14...Nf8, the position was:

click for larger view

15. a4

15. c5 sure looks tempting here:

"15. c5 would have been positionally more jjstified, shutting in the bishop and creating tension favorable for White on the queen-side,when it would have been extremely difficult for Black to create any activity. But Reshevsky has another plan.

15... Qc7

This move was cited as proof that Keres should have played 13...Qc7 to protect the e-pawn before playing Ne6 rather than losing time with 15...Qe7:

But these critiques all assume that 15...Ne6 immediately was impossible. But was it? If 15...Ne6 16. Nxe5 Black gets fine counter-play with 16...c5! e.g., 17. Nd5 NxN 18. exN BxN 19. dxN Qxe6 20. bxc5 Bc7 21. cxb6 axb6 22. c5 Be4 23. Bd3 Bxg2! 24. cxb6 (24. KxB leads to a draw after 24...Qg4+ 25. Kf1 Qh3+ 26. Ke1 Bf4 27. cxb6 RxB+! 28. fxR Qxe3+ and draws by perpetual check) Bxb6 25. Bf5! gxB 26. RxB Be4 with excellent drawing chances given the bishops of opposite colors.

None of the commentators appear to have considered 15...Ne6.

Perhaps best for Black, however, is 15...Ng4, a move some of the commentators considered but rejected (mistakenly in my view):

Golombek, in rejecting 15...Ng4, only considered the doubtful 16. Bc1 Red8 and 16. Bd2 (which he called better but did not analyze further). Keres did look at 15...Ng4 16. Bd2 but concluded that after the indicated 16...Ne6 Black has problems after 17. h3. In so concluding, Keres missed the lovely shot 17...Nd4! after which the advantage is with Black.

Both Keres and Golombek overlooked White's best reply to 15...Ng4: 16. Bg5. But even then, 16...Nf6 17. c5, while giving White some chances, is better for Black than the move Keres chose at the board: 15...Qc7.

Keres said that 15...Nh5 and 15...N6d7 should be considered, but Black seems to be in trouble after these moves if White responds forcefully with 16. c5.

All in all 15...Ng4 was the best chance for Black.

After Keres' 15...Qc7, the position was:

click for larger view

16. b5

"According to plan. White's advantage gains reality." (Kmoch)

As will be seen, however, Keres could have been OK after the text. Stronger was the immediate 16. c5. This move was analyzed and rejected by both Kasparov and Golombek.

"With the pawn on a4, the advance 16. c5 wouldn have involved the risk of this pawn being lost: 16. c5 Ne6 17. h3 bxc5 18, bxc5 Nd7." (Kasparov)

But after 16. c5 Ne6 (not Black's best--see below) White would have a clear advantage with 17. cxb6, e.g., 17...axb6 18. a5 !

Golombek found a better reply for Black, i.e., 16...Ng4, but his analysis also had holes:

"16. c5 fails against 16...Ng4 17. Bc1 [17. Bd2 is better, but White also has the edge with Golombek's 17. Bc1--KEG] bxc5 [very weak, far better is 17...Ne6 after which White's edge would be small] 18. bxg5 [missing the stronger 18. Ng5] Red8 [this looks like a blunder and seems to lose to 19. Bc4, but even with the superior 18...Nf6 White is definitely better].

In short, 16. c5 was almost certainly best.

After Reshevsky's actual 16. b5, the position was:

click for larger view

Time pressure was becoming a factor. Even before Reshevsky's 16. b5, the clocks read:

Reshevsky 1:26
Keres: 1:15

If the players from here didn't always find the ideal moves, we must remember that for them the clock was ticking (while I was able to devote the better part of two weeks studying this game with computer assistance whenever I felt I needed it.

Keres after 16. b5 played:

16... Red8

None of the commentators was happy with Black's situation after this effort:

"The choice of moves is unattractive." (Kmoch)

"With the text-move, Black tries to simplify the position somewhat by exchanges on the d-file." (Keres)

"Keres'' insipid move is indicative of his dissatisfaction with the development of events. He allows the opponent to develop his offensive even further."

So what to do? The commentators had three possible alternatives. I discuss these in ascending order of merit, and then add my own suggestion (16...Ng4)

(A) 16...cxb5

click for larger view

Very weak. Of course, if White responded 17. cxb5?? he would get crushed by 17...Bxe4! And if 17. axb5 Black would be fine after 17...Ne6 (better than Kmoch's N6d7). But the party would be over for Black if White played 17. Nxb5! Then, after 17...Qc6, Black gets killed by 18. Ng5 (rather than Kmoch's weaker 18. Nd6). This would have been far worse for Keres than his actual 16...Red8.

(B) 16... c5

click for larger view

Euwe's idea. But after 17. Nd5! Black is not much better than in the game. And Euwe's suggested follow-up of 17...BxN? was awful [Black has to play the unsavory 17...Qb8--KEG] and leads to a likely lost game after 18. cxB Red8 (Euwe's analysis ends here) 19. Nd2 (or 19. a5) Ne8 20. Nc4 (or 20. a5).

(C) 16... Ne6

click for larger view

This move, which in fact is far better than (A) and (B) above and also far better than Keres' actual 16...Red8, was rejected by Keres:

"Due to the earlier losses of time, Black is no longer able to carry out his main plan of 16...Ne6, because this would be answered by 17. bxc6 Bxc6 18. Nd5 BxN 19. exB Nc5 20. a5 and White has the better game thanks to his Bishop pair and attacking chances on the Queen-side." (Keres)

But Kasparov pointed out that after 20. a5 in Keres' line Black could play 20...Ng4; or if 20. h3 Nfe4; after which "Black would ave a quite acceptable position." (Kasparov)

In the above line, White would retain some edge with 20. Nd2 (instead of 20. a5), but this does not detract from the conclusion that 16...Ne6 was a major improvement on Keres' actual move and would have given him an entirely playable position.

But best for Black was:

(D) 16...Ng4! (the choice of Fritz and Stockfish but not mentioned by any of the commentators):

click for larger view

Play could then have become most exciting: 16...Ng4 17. Bg5 Ne6 18. a5 cxb5 19. h3! Nd4 (the melodramatic 19...Nxf2?! also seems playable) 20. axb6! axb6 21. Nb5 (still leaving his own Queen en prise) NxN 22. RxB Nf6 23. BxN BxB 24. Rdb1 Be7 with approximately even chances in a still complicated position.

xxxxxxxxxxxxxxxxxxxxxxxxx

But let's return to what actually happened. Keres played:

16... Red8

This left:

click for larger view

17. RxR

The beginning of a strong plan that brought Reshevsky to the brink of victory. But 17. a5 immediately, or perhaps 17. c5, were likely even stronger.

17... RxR
18. a5

click for larger view

"!"--(Keres)(Kasparov)(Wade, Whiteley, Keene)

"A strong move by which the breakthrough is begun." (Kmoch)

"Threatening 19. axb5 axb5 20. bxc6 and 21. Bxb6." (Golombek)

"White could of course not play bxc6 followed by Nd5, neither here not on the previous move, as Black in this case could simply capture twice on d5." (Keres)

"He has not sufficient time to play 18. h3 to prevent Black's next move since then would come 18...Ne6 followed by Nd4." (Golombek).

18... Ng4

"As a consequence of White's vigorous assault, the situation on Black's Queenside has become most critical." (Golombek)

"Black decides to try to solve the problems in the position by tactical means, however this is not the most successful decision when we take the approaching time trouble into consideration...After the text-move the position becomes very complicated." (Keres)

Since the text was ideal neither theoretically ("not much use"--Euwe) nor practically a good choice given the time remaining on the clock, let's examine the alternatives:

(A) 18...bxa5?

This can be disposed of quickly. As Golombek, Kmoch, and Wade-Whiteley-Keene point out, 18...bxa5 loses immediately to 19. b6!"

(B) 18...c5

This was Keres' suggestion, but as Golombek and Kmoch pointed out, this looks bad for Black after 19. Nd5.

Keres claimed that Black in this line could achieve "a defensive position without weaknesses" in which White would have "great difficulties obtaining any real advantage" via 19...BxN 20. cxB Ne8. But Kasparov looked deeper and pointed out that Black would in fact be in deep trouble after 21. Nd2 Nd6 [21...Nd7 might be slightly better, but Black would still have problems after 22. Nc4--KEG] 23. Nc4 after which "White can unhurriedly prepare the advantageous opening of the a-file." And White's edge would be even greater with 22. Bg5.

While Black can put up stiffer resistance with 19....NxN instead of 19...BxN, all the chances remain with White after 20. cxN after which Black is badly constricted and White has a protected passed pawn on d5.

C) 18...cxb5

Only Golombek mentioned this possibility, which looks better than the text and also better than (A) or (B). Golombek only considers the response 19. axb6, after which Black's position is not so bad: 19...axb6 20. Rxb5 N8d7 21. Nd5 NxN 22. exN (though Golombek--erroneously in my view-- says White's edge here is "marked"). But White can improve with 19. Nxb5 after which White's advantage can indeed be termed "marked," but not as severe an edge as with the moves discussed above.

(D) Black's most tenacious resistance seems to reside in 18...N8d7, a move none of the commentators consider (though White is still better with 19. axb6 axb6 20. bxc6 Bxc6 21. Ne1 with f3, Qd2, and Nd5 to follow.

After Keres' actual 18...Ng4, the position was:

click for larger view

Not a position I would fancy trying to hold in time pressure, even if it is not theoretically lost.

19. axb6 axb6

20. Bg5

"!"--(<babacova>)

As <babacova> has pointed out, with this move Reshevsky nicely followed up his aggressive play on the Queen-side by switching his focus to the other wing.

click for larger view

20... f6?

"?"--(Keres)(Kasparov)

"A mistake that shuts Black's Bishop completely out of the game, and weakens the important a2..g8 diagonal." (Keres)

"A serious positional mistake, which places Black on the verge of defeat. Now his bishop is shut out for a long time,but, more important, his king's defences are seriously weakened." (Kasparov)

That the text was bad seems clear. The more difficult question is what Black should have tried.

The commentators were unanimous that 20...Bf6 was correct.

"Better is 20...Bf6 hoping to eliminate White's powerful Bishop and, at the same time, keep the King-side intact." (Kmoch)

"20...Bf6 would definitely have been better." (Euwe)

"20...Bf6 had to be played..." (Keres)

I disagree.

Putting aside Golombek, who would have White play the clearly inferior 21. Bd2 after which Black has a difficult but probably holdable game with 21...Be7; the real question is how the proposed 20...Bf6 would hold up after 21. bxc6 (best) Bxc6.

Here, Keres has White playing the feeble 22. Bxb after which Black equalizes fairly easily with 22...NxB.

Kasparov gives the much stronger 22. Bd2 and nonetheless also reaches a decent position for Black, but only with (from Kasparov especially) some very sloppy analysis: His line from here is 22. Bd2 Be7 23. h3 Nf6.

So far so good, but then Kasparov gives 24. Nd5 BxN 25. exB N8d7 26. Bc3 leaving White with only a "slight advantage." While I think White's edge here is more than "slight," this is all irrelevant, since 24. Nd5? was a blunder and after 24...BxN 25. exB Black could turn the tables on White and get the better game with 25...Nxd5 (since the c-pawn is pinned). Instead, White should play--in Kasparov's variation--24. Bd3 with advantage.

But the more serious flaw with 20...Bf6 is that after 21. bxc6 Bxc6 White, rather than playing Kasparov's 22. Bd2 (and certainly rather than Keres' wimpy 22. BxN) could make real trouble for Black with 22. Qb3! Black then has nothing better than 22...Nd7 after which White continues with 23. Bc1! followed by Ng5 and Nd5. I'm not saying this is necessarily a win for White, but Black would have a long difficult defense ahead of him to hold the game.

The best move after Reshevsky's excellent 20. Bg5! was 20...Nf6. If then 21. bxc6 Bxc6 22. Qb2 (best), Black would have reasonable chances to hold on after 22...Rb8 23. BxN BxB 24. Nd5 BxN 25. cxB Nx7 or perhaps even better 22...N8d7 23. Nd5 Qd6 24. Rd1 BxN 25. cxB h6 26. BxN (the best try) NxB .

After Keres' weak 20...f6?, the position was:

click for larger view

Euwe says that Reshevsky here had a "slight advantage." All of the other commentators agree with Kasparov that Keres was, at best, "on the verge of defeat."

21. bxc6

"!"--(Kmoch)

"It is not necessary to prevent Black from blockading the position with c5." (Golombek).

21... Bxc6

"If 21...fxB? White wins by 22. Nd5 (if 22...RxN 23. cxR)." (Kmoch).

If--in Kmoch's variation--23...Bc8 24. Qb3 is brutal.

22. Bd2

"White threatens...23. Nd5...[The alternative] 22. h3 fxB 23. hxN Ne6 would only help Black." (Golombek).

After Reshevsky's very strong 22. Bd2, the position was:

click for larger view

"Now White has a great advantage owing to the pressure he can exercise on Black's Queen Knight Pawn. In combination with this, the command of d5 is of considerable importance (it would otherwise not mean much as the Queen Bishop file is open for Black).

22... f5?!

"Quiet play offer little chance to save the game so Keres tries a counter-action." (Kmoch)

"Black now has to continue aggressively..." (Keres)

Since the text creates an immediate crisis for Black, was there a less drastic solution? The only plausible candidate was 22...Ne6. The commentators all concluded that White eventually wins after 22...Ne6 though disagreeing about the best line for White in that event.

Keres gives 23. Qb3 as the winning line for White after 22...Ne6, but this was refuted by Kasparov:

23. Qb3 Nc5 24. Qxb6 QxQ 25. RxQ and now instead of Keres' 25...Nxe4? 26. RxB NxB 27. NxN RxN 28. BxN f5 after which White wins [but with 29. Rc8+ Kf7 30. Rc7+ Kg8 31. Bxf5 (31. g3 also wins) gxB 32. g3 and not with Keres' 29. h3 after which Black survives with 29...e4 rather than Keres' fxB? 30. hxg4] but with 25...Bxe4! 26. Nd5 Bd3 27. Bf1 BxB 28. KxB and now 28...e4! (superior to Kasparov's 28...Nd3, which also probably is sufficient to hold the position).

Instead, after 22...Ne6 White should play Golombek's 23. h3! also favored by Kasparov: 23. h3. Nh6 24. Qb3 Nc5 25. Qxb6 QxQ 26. RxQ and now:

26...Nxe4 as given by Golombek and Kasparov which loses not to Golombek's 27. NxN but to Kasparov's 27. RxB! NxB 28. NxN RxN 29. c5! Rc2 30. Bc4+ Nf7 (of course not 30...Kh8 31. Rc8+ Bf8 32. RxB+ Kg7 33. c6 or 30...Kf8 31. Rc8+ Ke7 32. Nd5+ Kf7 33. Ne3+) 31. Rc7 Rc1+ (Kasparov's 31...RxN loses immediately to 32. BxN+) 32. Kh2 RxN 33. BxN+ Kh8 34. Bd5 and White's c-pawn brings him victory.

In sum, the theoretically most determined resistance was 22...Ne6, but since White even then will almost certainly win, Keres' 22...f5?! was a good practical. All in all, I really like this move, much as Fritz and Stockfish condemn it.

After 22...f5?!, the position was:

click for larger view

23. Bg5

"?!"--(Kasparov)

"Avoiding Black's little trap: 23. exf5 e4 24. Nxe4 25. BxN 26. QxB RxB where White loses piece." (Kmoch)

Actually, White still has the better chances on Kmoch's line after 27. fxg6 Kh8 (if 27...hxg6 28. c5 RxN [forced] 29. QxR Qxc5 30. Rb5) 28. Re1 Rd4 29. Qc2 Rd8 29. h3 Ne5 30. gxf7 or 30. NxN, though a draw seems most likely in either case. So 23. exf5 can't be right, even if it is not a losing move as Kmoch and Golombek thought.

But was there anything better than Reshevsky's 23. Bg5:

<Honza Cervenka> and Kasparov correctly claim a win for White after 23. Qb3 Kh8 (23...fxe4 loses to Kasparov's 24. c5+ Kh8 25. Ng5 Nh6 26. Qxb6 [26. cxb6 is even more deadly--KEG]) 24. Nd5 ("a pretty straightforward win for White"--<Honza Cervenka>) and then--as per Kasparov--24...BxN 25. cxB Qc5 (25...Nf6 or 25...h3 also lose) 26. Be1 (and even more simply, 26. Qxb6--KEG).

Thus, Reshevsky had a win for 23. Qb3.

But in fact Reshevsky's move (23. Bg5) also wins. Given that Reshevsky was facing serious time pressure, the fact that there may have been a somewhat more crushing move hardly detracts from Reshevsky's achievement here. As will be seen, however, as the time pressure for both players became more intense, both sides missed the theoretically best moves, and Reshevsky likely gave Keres a couple of theoretical chances to save the game.

In any event, after 23. Bg5, the position was:

click for larger view

23... Re8

"Here 23...Bf6 was bad due to 24. h3 [even better would be 24. exf5--KEG] BxB 25. hxN...The text-move is necessary to be able to answer 24. exf5 with 24...e4." (Keres)

After 23...Re8 the position was:

click for larger view

Meanwhile, the clock was looking scary for both sides:

Reshevsky seems an extremely crafty attacker if he has even a slight advantage and he sees or senses potential opportunity. Give him a tempo or two and forget about it. Feels like he's the epitome of a player "on a roll" or " in the flow" when he puts one of these attacks through.

The fact that Reshevsky never became world champion is no major blot on his memory. He never had a chance to play for the title before this 1948 tournament, which was his only real chance to win the title. Reshevsky sat out the 1949-1951 and 1957-1960 cycles for non-chess reasons, in the interim he had to confront Smyslov as Smyslov's peak, and after 1960 Spassky, Fischer, and Korchnoi (who demolished Reshevsky in their 1968 match) were on the scene and were almost certainly stronger players.

Indeed not, same as there were many fine NBA players who never got a ring in the 1960s (or only one) due to facing a team which won everything in sight.

24. h3

As Golombek and later Kasparov pointed out, 24. exf5? would be a "terrible blunder" because of 24...e5! (25. Ne5 QxN 26. BxN Qxc3 27. QxQ BxQ 28. fxg6 Nxg6).

Similarly, <rilkefan> and <perfidious> have refuted 24. Nd5 which loses (as they have shown) to 24...BxN 25. exB e4.

The consensus seems to be that Reshevsky had to drive back Keres' g4 Knight. But whereas White's win remained uncertain after the text, it would have been less problematic with the real killer: 24. c5! (a move not discussed in any of the voluminous literature I have seen on this game).

After 24. c5, Black pretty much has to play 24...bxc5. Then White wins with 25. Nb5 Qb8 (or 25...BxN 26. BxB Rb8 27. exf5 [27. Bc4+ is also a crusher] Kh8 28. Bc4 h6 29. RxR QxR 30. h3 Nf6 31. BxN BxB 32. fxg6) 26. Nd2 Ne6 27. Bc4 Kh8 28. BxN RxB 29. Nd4 after which Black would have to give up his Queen to have any chance: 29...QxR+ 30. QxQ exN 31. exf5 gxf5 32. Qxf5 after which White should win easily.

But Reshevsky's move still left him in the driver's seat (albeit with a less certain win, the position after 24. h3 being:

click for larger view

24... fxe4

"?"--(Euwe)

This move by Keres has come in for criticism, all of which strikes me as unfair.

Let's hear from the nay-sayers:

"This [the text] only aggravates the situation, since it helps Black's opponent secure the Bishop pair." (Euwe)

"In time-trouble, Black does not have time to calculate all the complicated lines in the position and therefore chooses a simple continuation, but it is clearly better for White." (Keres)

Most commentators content that Keres would have had better chances with 24...Nh6. But Kasparov rightly disagreed, and stated that:

"Keres was wrong to criticize his move, which enabled him to reduce the effect of the knight invasion at d5, simplify the position, and thereby ease his burden a little, although White's advantage is still tangible."

The touted alternative, 24...Nh6 was in fact much worse. White has at least two ways to win (though not necessarily with 25. BxN as suggested by Keres and Kasparov, after which Black would have some chances to survive with 25...BxB 26. exf5 [26. Bd3, not mentioned by Keres or Kasparov was best but hardly a clear win] e4 27. Nd4 e3 28. f3 [Kasparov's move, 28. f4 would be better but hardly a win] Qg3! [and not Kasparov's 28...Bb7] and Black is certainly not worse let alone losing):

(A) 25. Be3 (recommended by Golombek and Euwe) Nd7 26. Qb3 and now if 26...Kh8 27. Nd5 Qd6 28. Rd1; and

(B) 25. c5! (not mentioned by any of the commentators but clearly best, e.g., Nxf2! (the best chance) 26. KxN Nxc5 27. Kg1.

After Keres' quite reasonable 24...fxe4, the position was:

click for larger view

25. Nxe4?

This second best move by Reshevsky, which gave Keres real chances, has passed without comment. But correct and winning here was 25. Nd2! If now 25...Ne6 (best) White wins with 26. c5 NxB 27. cxb6! Qd6 28. BxN e3 29. fxe3 Qc5 30. Qa2+ (or 30. Qb3+) Kh8 31. Nc4 leaving White up a pawn with a powerful advanced passed b-pawn; if 25...Nf6 26. Qb3 Ne6 27. BxN Nd4 (or 27...BxB 28. Ndxe4 Bd8 29. c5!) 28. Qxb6 QxQ 29. RxQ BxB 30. Bg4.

Make no mistake, Reshevsky still was much better, but the text gave Keres more than a ghost of a chance.

After 25. Nxe4?, the position was:

click for larger view

25... BxN

"Otherwise White forces the exchange of the other Bishop, enfeebling Black's King position." (Kmoch)

If instead 25...Nh6 (which Keres said was "also possible" White wins not with Golombek's 26. Nf6+ (26...BxN 27. BxB Nd7 [much better than Golombek's 27...Re6] or Keres' 26. BxN (26...BxN 27. QxB BxB with reasonable chances for Black) but with 26. c5! BxN (not 26...bxc5? 27. BxN BxB 28. Nf6+) 27. Qxe4 Qxc5 28. Be3 with Rxb6 and Rb7 to follow.

26. QxB

"Two Bishops means a further increase in White's advantage. Reshevsky has repelled the counter-attack and now starts to realize his assets." (Kmoch).

26... Nf6

click for larger view

Resheveky: 2:14
Keres: 2:15

27. Qe3 N8d7

27...Ne6 was perhaps slightly better, but Keres' move was reasonable objectively and excellent given his horrendous time pressure.

28. Qb3

click for larger view

28... Rb8

While not necessarily fatal, this move was clearly not best. Only <goodevans> on this site has stated the obvious:

"Only Qb3 had been played, my instinct would have been to get my king off the a2...g8 diagonal."

Quite right. Thus, 28...Kh8 was best.

Reshevsky: 2:21
Keres: 2:24.

29. Be3

"Threatening to win a pawn by 30. c5+." (Kmoch)

"...a cute positional move which prepares Ng5 and puts pressure on black's weakened queenside." (<babakova>).

Well put!

The text was better than 29. c5+ immediately:

"Naturally not 29. c5+ Kh8 and in the end White would only achieve the exchange of the c4 and b6 pawns, which would free Black from all his problems." (Keres) (see Golombek, similarly).

After 29. Be3, the position was:

click for larger view

29... Nc5

Hard to believe that Keres yet again failed to play 29...Kh8 and that none of the commentators faulted him for this.

The only commentator even to mention 29...Kh8 was Golombek, but only to criticize it:

"If 29...Kh8 30. Ng5!"

True, but Black should have faced even more serious problems after the text, the position now being:

click for larger view

30. Qc2

"?"--(Reshevsky)

This seemingly appropriate move, which was otherwise not commented upon by the analysts who considered the game, elicited the following amazing note from Kasparov:

"Reshevsky conceives an original, but not the most effective plan. Meanwhile, by the paradoxical exchange of his key bishop at e3 for the knight at c5, Whjite could have decisively exploited the weakening of the a2-g8 diagonal created by Black on the 20th move: 30. BxN! QxB 31. Ng5 Kh8 32. Ne6! (32. Qg3?! leaves White with only an extra pawn in the endgame...) 32...Qc8 33. c5!, beginning an elegant attack."

I have quoted only part of Kasparov's amazing analysis (which in some variations runs 13 moves deep). Kasparov acknowledges that his analysis came from his "silicon monster," and my little monsters reached the same conclusion. I highly recommend reading Kasparov's analysis in his book My Great Predecssors, Volume IV, p. 55.

I think it unfair to affix a "?" to Reshevsky's move. Few if any humans would play 30. BxN here (it looked ludicrous to me at first blush) and to expect Reshevsky in real time (and in time trouble to boot) to unearth the gem Kasparov discovered with computer assistance is just not practical.

Kasparov's wonderful variations are worth reporting, since they undoubtedly represent the theoretical "truth" concerning the position. All I can say is BRAVO, and then return to real life, the position after 30. Qc2 being:

click for larger view

Given the severe time pressure, the play of Reshevsky and Keres until Keres' weakening beginning on move 35 was highly commendable (and in my view outstanding) even though several commentators (including Keres himself) accused the contestants of engaging in "aimless maneuvering."

30... Ra8

"Better 30...Re8 at once, but Black has only five minutes left for ten moves; hence the purposeless Rook move." (Golombek)

"...taking into consideration that the players now faced serious time-trouble, the following aimless maneuvering, especially from Black's side, is somewhat understandable." (Keres)

But is 30...Ra8 in any way a poor move? It was the first choice of Fritz and Stockfish! The fact is that Black has little plausible chance for counter-play and can only sit tight and await White's attack. I see no reason for Keres to have apologized for this move, nor was 30...Re8 any real improvement.

31. Ng5

The beginning of Reshevsky's planned attack.

Keres suggested that 31. Nd2 was stronger, but I don't see how that advances White's cause more than the text. On this point, I agree with Golombek:

"Preparing to gain further command of the board by Bf3, and at the same time having in mind an ingenious scheme for a K side attacj containing hidden threats on Black's h7."

31.... Re8

"So as to meet 32. Bf3 with e4." (Golombek)

He might have played e4 here, but the text was not so bad, despite Euwe's comment:

"Black does not know exactly where to go."

I would have called this unfair to Keres--except that Keres agrees!!

In any case, after 31...Re8 the position was:

click for larger view

Now things really got exciting:

32. h4!

"!"--(Golombek)(Kmoch)(<babacova>)

"Suddenly, out of the blue, Reshevsky conjures up a king side attack." (Golombek)

"Very strong and deep. Threatening 33. h5, White provokes e4 and opens the way to d5 for his Knight." (Kmoch)

"The move I like the most [in this game] is 32. h4! which is incredibly profound. It covers g5 an extra time and prepares to swing the Knight to f4 [via h3]." (<babacova>).

As for the alternative, 32. Bf3, as Kasparov has demonstrated, Black could then reach a likely drawn Bishop of opposite colors position via: 32...e4 33. BxN [the Bishop for Knight trade theme again!--KEG] bxB 34. Nxe4 NxN 35. BxN Bd4.

All in all, 32. h4! was part of a fine conception by Reshevsky, leaving the position as:

click for larger view

32... e4!

"This move, which limits the scope of White's light-squared bishop, is sooner or later forced. The pawn on e4 is so well protected that Black need not fear losing it now." (Keres)

As Golombek noted, 32...h6? would have been a disaster in light of 33. Qxg6 hxN 34. hxg5 after which Black might as well resign.

32...e4 left the position as:

click for larger view

And the clocks now read:

Reshevsky: 2:27
Keres: 2:27

Pretty scary for both sides in this extremely difficult position.

33. Nh3

This move shows one of the keys to 32. h4: allowing the Knight to get to f4 via h3.

33... Rd8

"?"--(Keres)(Kasparov)

Keres and Kasparov both believed the Rook was badly posted here and that Black's best chance lay in 33...Bf8.

Keres said that after 33...Bf8 he would have been able to respond to 34. Nf4 with Bd6. In fact, White had a stronger option (see below) and if indeed 34. Nf4 best for Black would be 34...Bh6 or 34...Ncd7, in either case leaving White with excellent chances.

Kasparov identified a better response to 33...Bf8, i.e., 34. Qb2. He gives the resulting line: 34...Re6 35. Nf4 Rd6, all of which leaves Black no better off than the text.

The best chance for Black here was probably 33...Qd6, but even then White would be strongly placed after 34. Rd1.

The simple fact is that Keres had a difficult position (and was in desperate time trouble). His move was not noticeably worse than the alternatives.

34. Nf4

All according to plan.

click for larger view

"Threatening to win the exchange by 35. BxN and 36. Ne6." (Golombek)

34... Rd6

"Due to the threat of 35. BxN followed by Ne6, Black no longer has time to prevent the white knight from invading on d5." (Keres)

"Black's position is again very difficult..." (Kasparov)

This move certainly left Keres poorly placed, but the alternatives were not much more savory.

Keres recommended 34...Re8. This might provide some small relief if White followed Golombek's suggestion and played 35. Rb5 (e.g., 35...Bh6 36. Nh5 NxN 37. BxB Nf4, though White could just play 36. g3 and build up the pressure slowly), but Keres' 35. Nd5 would have presented more severe problems e.g., (following Keres' line): 35...NxN 36. cxN Be5 but then instead of Keres' "pawn grab" with 37. BxN White could play 37. Qc4 with all sorts of nasty threats.

Probably Black's best try here would be 34...Bh6. Then, after Kasparov's suggested 35. Ne6 NxN 36. BxB Nd4 37. Qb2 NxB+ 38. QxN Black has descent chances to survive with either 38...Re8 or 38...Rd4.

The strongest line for White after 34...Bh6 is Kasparov's alternate suggestion of 35. g3! BxN 36. BxB with some basis for hope after 33...Qc6.

Stockfish selects 34...Qc6 from the poor options open to Black, but this allows White to build his attack slowly but surely with 35. Qa2 and 36. g3

Given the extreme difficulty of his position, and especially considering his time trouble, Keres' 34...Rd6 (which was certainly not much worse than the other possibilities) can hardly be faulted.

After 34...Rd6, the position was:

click for larger view

35. h5!

"!"--(Golombek)(Kmoch)(Euwe)

"After this further loosening of his position, Black is helpless." (Golombek)

"Decisive." (Euwe)

But was 35. h5 "decisive" or indeed even best.

Keres said it was not:

"A typical bluff in time trouble, which Black however falls for."

Keres goes on to say that 35. Nd5 was best for White and gave good winning chances. He gives the following line:

35. Nd5 NxN 36. cxN Be5

This would leave:

click for larger view

Keres here considered two possible moves for White:

A) 37. BxN QxB (37...Bd4 may be slightly better for Black here) 38. Qxe4 Qd4 and claims that the resulting opposite bishop ending would be drawish. But there are several problems with this analysis. First, as Kasparov pointed out, White in this line would have very real winning chances with 39. Rb4. But, more fundamentally, Black would probably be able to hold the game with 38...Bd4. Thus, 35. Nd5 followed by 37. BxN is inferior to Reshevsky's 35.h5.

B) 37. Bg4 was the move Keres said was best "with a better position, but its' not without some chances for Black." In fact, however, Black would likely be able to hold with 37...Kg2 38. Qa2 Qf7 39. Rd1 h5

C) 37. h5. This is the critical line after 35. Nd5. White is strongly placed after 35...gxh5 38. Bxh5.

The real problem with 35. Nd5, however, was identified by Kasparov:

"Reshevsky was again guided by psychology in his choice of move. 35. Nd5 had one drawback--it was too obvious, where the non-concrete pawn sacrifice could have come as a surprise, which would be especially unpleasant in time-trouble."

in fact, the best alternative for White in a theoretical sense was Kasparov's 35. g3, after which depending on Black's response Reshevsky could have continued with 36. Nd5 of 36. Rb5 or perhaps even 36. h5.

In the context of Keres' (and Reshevsky's) time trouble, however, 35. h5 was almost certainly correct. As Kasparov said:

"...the sudden advance of the pawn costs the opponent precious time, since it forces him to solve concrete problems."

For this reason, I have affixed a "!" to 35. h5, which left the position as follows:

click for larger view

35... g5?

"?"--(Keres)(Kasparov)

Time pressure led Keres to weaken his King-side with fatal effect.

"This advance not only loses a pawn, but it also weakens the black kingside decisively, so that White, in addition to everything else, also gets a strong attack against the king." (Keres)

"The trick succeeded; Keres did not risk taking the pawn and made an altogether poor move." (Kasparov).Nxh5?

Since the text is a losing blunder, what should Keres have played? Let's examine the alternatives:

(A) We cam dispose of 35...Nxh5? which loses to Kasparov's 36. BxN gxB 37. Nd5!Qd8 38. BxN bxB 39. Qxe4 ["...and this position cannot be held" (Kasparov)], or even better and simpler 36. NxN! gxN 37/ Bf4 and Black is busted.

(B) 35...Qf7 (Kasparov's second choice), which--though better than Keres' actual move-- loses to both 36. Ra1 Qe8 37. hxg6 hxg6 38. Ra7 Rd7 39. Ra3 Kf7 40. Qb1 Qd8 41. Nd5 NxN 42. cxN Be5 43. BxN bxb 44. Qe4 and White, with a pawn plus as well as the attack, should win and to 36. Nd5 NxN 37. cxN Be5 38. BxN bxB 39. Qxe4.

(C) 35...Bh6. a move not mentioned by any of the commentators, is also better than Keres' choice but also ultimately insufficient: 36. Nd5 NxN 37. BxB Ne6 38. Qa4 Nf6 39. Be3 Nd4 40. BxN RxB 41. Qb3 Qc5 42. hxg6 hxg6 43. Qxb6 QxQ 44. RxQ leaving White a pawn up in a likely winning ending.

(D) 35...gxh5 This is by far the best choice and gives Black at least a fighting chance to survive. 36. Nd5 (not 36. Nxh5 NxN 37. BxN Bh6 [much better than Golombek's 37...Re6 which loses to 38. Bg4 or 38. Qd1]) and Black should be able to hold) 36...NxN 37. cxN which would leave the position as follows:

click for larger view

Here, Keres and Kasparov suggested differing was to save the game:

DI) 37...Bh6 (Keres' move). In response, Keres and Kasparov only consider the weaker 38. BxN after which Black can hang on with 38...e3! (much better than 38...QxB discussed by Keres and Kasparov) , e.g., 39. Rxb6 (best) exf2+ 40. Kf1 RxR 41. d6 (forced) Rb1+! 42. QxR QxB 43. d7 Qa5 and Black should hold. But the crucial line in this variation (ignored by both Keres and Kasparov) is 38. Bd4! after which Black would have to work hard to hang on, e.g., 38...Bg7 (forced) 39. BxN QxB 40. Qxe4 Rf6! and Black has decent chances to survive based on the opposite color bishops and his pressure on f2.

In sum, Keres' 37...Bh6 probably holds.

DII) 37...Be5 38. Bxh5 (much better than Kasparov's 38. BxN QxB [not Kasparov's 38...bxB] 39. Qxe4 Qd4 40. Qf5 (better than Kasparov's 40. Rb4) Rxd5 41. Qe6+ Kh8 42. Rxb6 Rd8 and Black should be able to hang on) Kh8 39. Qc4 with a tough fight ahead but with very real drawing chances for Black.

In sum, 35...gxh5 was best for Black and left Black with reasonable chances to hold the game.

Needless to say, Keres--with just seconds left on the clock--has no chance to investigate all of the above.

In practice, Keres played 35...g5?, leaving the position as:

click for larger view

Keres was now toast:

36. Nd5

"!"--(Kasparov)(Kmoch)

"To conquer and occupy effectively this square has been White's aim from the beginning. Now he achieves all he wants. Black must lost at least a Pawn in a very bad position." (Kmoch)

"The triumph of the consistent play for seizure of the d5-square)." (Kasparov)

36... NxN
37. cxN

click for larger view

37... h6

"?"--(Kasparov)

"This conclusively weakens the light squares, hastening the inevitable end. However, it is no longer possible to offer Black any good advice." (Kasparov).

Unsavory as the text is for Black, there was nothing significantly better. So Kasparov's "?" seems unduly harsh.

Euwe said that 37...Kh8 was "slightly better." But after 38. Bxg5 [much stronger than Golombek's 38. BxN] Qf7 39. Qc4 Qxd5 40. Be7 QxQ 41. BxQ Rc6 42. Bd5 Rc7 43. Bc8 Rc8 44. Rxb6.

37...Bh6 (which Kasparov says was recommended by Keres [though I don't see this anywhere in Keres' notes on the game] gets crushed not so much by Kasparov's 38. BxN (which, in fairness, also wins) but by 38. Qa2 (or perhaps 38. Qc4) Qc8 39. Qa7 Qd7 40. Qb8+ Bf8 41. Bxg5.

I initially thought 37...Bf6 was an improvement, but according to Fritz it loses to 38. Rb4 e.g., Bd8 39. Rb5 Qe7 40. Bd4 h6 41. g3 Qe8 42. BxN bxB 43. Qxc5 and Black is done for.

In short, Keres had no saving move and his 37...h6 was no worse than the alternatives. it left the position as:

click for larger view

38. BxN

"!"--(Kasparov)(Keres)(Kmoch)(Wade-Whiteley-Keene)

"White liquidates: but his main aim is not to win the pawn, but the greater elbow room that the white pieces will get after the text move. This will enable him to launch an immediate mating attack." (Euwe)

"Winning a pawn and assuring White of a strong attack against the king. The presence of opposite-colored bishops is only in White's favor, as it facilities the conduct of his attack." (Keres)

In fact, the text was not best. As will be seen, Black (if not in awful time trouble) could now have made White's task much harder (though the game is still won for White). The strongest winning move here (as confirmed by Fritz and Stockfish) is 38. Rb4. e.g., 38...Be5 39. Rxe4 etc.

After 38. BxN, the position was:

click for larger view

38... bxB?

As Keres pointed out after the game, 38...QxB would have been better, e.g., 39. Qxe4 Qd4 (much better than Golombek's disastrous 39...Qd4 40. Qe8+ Kg7 41. Qe7+ Kg8 42. Bg4 etc. and also better than Keres' 39...Rf6 40. Rf1 Bf8 41. Bc4 b5 42. Bd3) 40. Qe8+ Bf8 41. Rd1 Qb4 after which Black can resist for at least a while.

39. Qxe4

click for larger view

39... Bd4

"?"--(everybody)

"Losing at once" (Golombek)

"This loses at once. It is true that there was no doubt about the game's [result]..." (Kmoch)

"All moves would have lost,but this one makes exceptionally short work of it." (Euwe).

"The last mistake, in severe time-trouble; it loses on the spot." (Keres)

"This loses immediately." (Kasparov)

Golombek said that Black "would have lasted longer after 39...Kf8." But he only considered 40. Rb3, whereas 40. Bg4 (en route to e6) would have been a killer.

Golombek also gave 39...Rd8 as a better option, but that would also get destroyed by 40. Qe6+ (much better than Golombek's 40. Bc4 as pointed out by Kasparov) Qf7 (continuing with Kasparov's line) 41. Rb7 QxQ 42. 42. dxQ Bf6 43. Bb5 Rd1+ 44. Kh2 Re1 (44...Rd4 was slightly better, but no true balm) 45. Bc4 Kh8 and now 46. f3 (better than Kasparov's 46. Rf7) and the game is over.

The stiffest resistance would have been offered by 39...Bf8 (a move mentioned on this site by <wwall>), but White would still win with 40. Rb5 (even stronger than 40. Bg4 as given by Kmoch) after which White's attack can not be held at bay, e.g., 40...Qe7 41. Qc4,

After 39...Bd4, the position was:

click for larger view

Reshevsky not finished off the game neatly:

40. Qe8+ Kg7
41. Rb8

click for larger view

1-0

"He will either be mated or lose his Queen." (Euwe)