I think <Eyal>'s comment about 33 Rxe4! is exactly right. It is easy to calculate, but hard to see.

In fact my own reaction to the 34 Rxe4 prompt from my computer was to blink three times and check I had the right position set up.

Only after a period of disbelief, I of course saw how "simple" it was.

It's actually Monokroussos' comment, but I also agree… that's a case where the basic ("geometrical") concept is the difficult thing, rather than the calculation - a rook sac is required in order to gain a tempo for transferring the queen from a5 to c7 via c3 (with check), while keeping in mind that the b1-h7 diagonal is denied from the black king because of Bc2. Topalov's notorious miss in the 2nd Elista game was actually of something considerably simpler:

click for larger view

Here 32.Rxg4+ & 33.Qc7 are both very straightforward moves, one only needs to pay attention to both sides of the board rather than focus only on the K-side (and the b1-h7 diagonal, as far as the queen in concerned).

There are many books and materials on the subject, but the general rule is that if the king on the side without the extra pawn can reach the queening square of the final pawn, the game is a draw. In the endgame here, black's king is well placed so white has no hope of cutting it off from potential queening squares once the other pawns are gobbled up.

This time it is less clear to me if Topa missed it or he simply discarded it. The computer line only exchanges bunch of pieces. After Qc7 there are at least to viable continuations for Aronian:

36.Qc7 Qb1+ 37.Kh2 Qxg6 38. Qxd8 Kg7

click for larger view

or 36.Qc7 Rd7 37.Bxd7 Nxd7 38.Qxd7 Qxg6

click for larger view

Didn't Tal say that when he was looking for a move, he first looked at what he could sacrifice.

Would an attacking player see 34 Rxe4!! easier than a positional player? It would be good to get a poll from some top players asking whether they would have seen it. (honestly)

Is it just a bit of slack form? You would think that a top player analyses every possible move, but for time consumption occasionally prunes his analytical tree. 33 Rxe4!! is an easy quick move to prune, because it loses a piece for no obvious recapture of material. If a player keeps disciplined and always looks 2 or 3 moves ahead in seemingly ridiculous combinations, then he will never miss these chances.

Topalov only had to look one more move ahead 35 Qc3+ and he would have seen the whole concept. But why bother keep looking at 34 Rxe4!!

<Blunderdome: Also, the rook has been pointing at e4 since move 29...hard to see when possibilities arise in what appears to be a static part of the position when all the moves are taking place on other parts of the board.>

We all try to keep our analysis disciplined, and try not to dismiss any move, but to keep dismissing the same move over and over again,(if that's what he did) it then must be difficult to all of a sudden bring the move back for consideration.

I forgive Topalov for not finding 34 Rxe4!!

BUT it's still a lesson to be learned. 'prune your tree with care'. Get into the habit of ALWAYS analysing the moves that also seem ridiculous. Easier said than done :-)

Missing a possible win is probably a 'massive deal' to Topalov, so I'm sure he'll be re-assessing his thought processes in time for the world championships. Something you have to do if you never want to miss moves like 34 Rxe4!!

<33... Nb6 Now all that's left is to demonstrate the brilliant blow 34. Rxe4!!

The key line is:
33... Nb6 34.Rxe4!! Qxe4 35.Qc3 Kg8 36.Qc7

It's all very simple. The main thing is to see the idea itself.>

Followed by:

<34.Bb3?? From someone else, perhaps, but I didn't expect such blindness from Topalov. It turns out that the sun also has spots.>

I think that in middelgame positions, as a rule one definitely wouldn't analyze every possible move, since that would indeed be suicidal in terms of time consumption (in the position after 33…Nb6, for example, there are 33 legal moves for White). Actually, in the case of top players I wouldn't be surprised if they often semi-consciously prune in advance a larger number of moves than weaker players, because they have strong tactical & strategic intuitions about what might "work" and should be focused on, and what might not. And that's exactly one of the points in which computers are so much stronger than human players - they don't play by "intuitions" or "plans," or consider some moves as less "natural" than others, but just check all possible moves each time without prejudice... it often means a huge waste of time, but it also means that they practically never miss "strange" moves that work tactically.

<polarmis: Apparently in an interview with the Spanish Chessbase after the game he said that he was already disheartened at letting an advantage slip by that point and simply wasn't optimistic about his position.>

<34.Bb3? Not even a second-rate move, but much weaker.

[Firstly, it was possible to achieve an unbelievably beautiful victory with 34.Rxe4!! Qxe4 35.Qc3+! Kg8 36.Qc7, and in order to avoid mate black has to give up a rook: 36...Qb1+ 37.Kh2 Qxg6 38.Qxd8+, and there's no need to go on.]

[Secondly, it was possible to pick off the whole queenside with: 34.Qxa7 Nxa4 35.Qxb7+ Kxg6 36.Qa6+ Kg5 37.Qxa4, retaining an extra pawn.]>

Vasiliev at Chesspro actually put the same thing to Topalov: http://chesspro.ru/_events/2010/lin...

<The fact that you didn't see Re4 isn't so terrible, it's not so easy to spot, but why didn't you take the pawn on a7 and win an elementary endgame with two connected passed pawns?

Veselin shakes his head as though trying to drive away a bad dream, makes a funny grimace as if to say why are you asking me about something that's absolutely impossible to explain? And then he wipes the smile from his face and says with great conviction in his voice:

Only after a period of disbelief, I of course saw how "simple" it was.>

Wow, an honest kibitzer here besides me. :-)

I read online that 34 e1xe4!! won on the spot, looked at this position, and *still* did not see the win, just like you.