1.e4 e5 2.Nf3 Nc6 3.Bc4 (<The Italian Opening is probably my first chess love as openings go, and as such, it's kind of hard to forget about her... I mean it.>) Bc5 4.b4 (<The good old Evans Gambit.>) Bxb4 (<The gambit can be declined with 4...Bb6, but as the old saying goes, the best way to refute a gambit is to accept it.>) 5.c3 Ba5 (<5...Be7 is another, calmer alternative.>) 6.d4 exd4 7.O-O Bxc3 (<I wouldn't refer to this a mistake as much as being very risky. It was only tried 2 times in the Opening Explorer .>) 8.Nxc3 dxc3 9.Qb3 (<Diverting from 9.Qd5 and 9.Ba3 in the database and trying to bring the game into the more common Compromised Defense variation, which goes 7...dxc3 8.Qb3 Qf6 9.e5 Qg6 10.Nxc3 Bxc3 (10...Nge7 is more common), but black variates again.>) Qe7 (<9...Qf6 is an alternative.>) 10.Qxc3 Qf6 11.e5 Qg6 12.Ba3 Nge7 13.Rad1 O-O 14.Rfe1 b6 15.Bd3

click for larger view

f5 (<? The decisive mistake, it just loses a piece. Black had to try 15...Qh6 (15...Qe6? 16.Bxh7) 16.Be4! Rb8 (white was threatening 17.Bxe7) 17.Bxe7 Nxe7 18.Qxc7 Ba6 19.Rxd7. During the game, I had intended to play the 16.Re4-Rh4 manoeuvre, which isn't as strong, though, with something like 16...Qe6 (16...Bb7? 17.Rh4) 17.Rh4 h6 and here I couldn't find a clear plan on how to continue. Maybe 18.Bc4 Qg6 19.h3 with the idea of 20.Rg4. In the last line, black can also try 18...Qf5 and even there 19.h3 might be the best move, with 20.g4 in mind.>) 16.exf6 Qxf6 17.Qxf6 Rxf6 18.Bxe7 Nxe7 19.Rxe7 Bb7 20.Ne5 d6 21.Rxc7 Bxg2 22.Bc4+ Kf8 (<22...d5 23.Rxd5.>) 23.Nd7+

(<The final position:

click for larger view

Had black found the correct defense on move 15, the game would certainly have lasted much longer and it might have gone either way. Still, it's not easy to see how black can unwind (especially since he's also trying to maintain his material advantage), so perhaps black's problems could be traced to as far as his 7th and 9th (especially in combination with 10...Qf6) move.>)

1-0

1.e4 c5 2.Nf3 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e5 6.Ndb5 d6 7.Bg5 a6 8.Na3 b5 9.Nd5 (<The other, also very popular and "wilder" alternative is 9.Bxf6 gxf6 10.Nd5 f5 (10...Bg7) 11.Bd3 (11.c3) Be6 etc.>) Be7 10.Bxf6 Bxf6 11.c4 (<11.c3 has been tried more often: Opening Explorer , for example, Anand vs Van Wely, 2006 went 11.c3 Bg5 12.Nc2 Rb8 13.a4 bxa4 14.Ncb4 Bd7 15.Bxa6 Nxb4 16.cxb4 O-O 17.O-O Bc6 18.Rxa4!. The Sveshnikov hasn't been as popular on top level as it used to be, especially thanks to Leko during his incredible run towards the Brissago final and Kramnik. Other recent games in the 11.c3 line: Shirov vs Topalov, 2008, Leko vs Carlsen, 2008, Anand vs Shirov, 2008 (all 1-0). >) b4 12.Nc2 a5 13.g3 Bg5 (<One of black's advantages is having the two bishops, also on g5 the bishop has better mobility.>) 14.Bg2 O-O 15.O-O Be6 16.f4 Bh6 (<According to the database, this is a new move. Z Bogut vs Filippov, 2006 went 16...Bf6 17.Nxf6 Qxf6 18.b3 etc., with an eventual draw>) 17.f5 Bxd5 (<I didn't like the alternative 17...Bd7 18.f6 (also preventing 18...Ne7, an idea behind 17...Bd7, preserving the two bishops) g6 as much, although black is still doing well.>) 18.exd5 Qb6+ (<A zwischenzug, aiming at simplifying the position a bit.>) 19.Kh1 Nd4 (<I know, it's a cowardly move, but I didn't see any other way to counter white's plans on the kingside. After 20.Nxd4 Qxd4 21.Qxd4 exd4 22.Rad1 Be3

click for larger view

click for larger view

(<White has obviously no desire to exchange everything on d4 as in the previous comment. However, this move has now given me a reasonable plan, something I didn't really have before.>)

Nxc2 22.Qxc2 Rfc8

(<So, my plan from now on will be undermining white's queenside with a5-a4 and after either abx3 or bxa4 penetrating along the a-file or c-file (depending o the situation) with the rooks and queen, trying to win white's pawns.>)

23.Be4 a4 24.h4 (<White had the choice of defending passively or trying to counter on the kingside.>) Qe3 (<! - The queen comes to a great square with tempo, attacking the g3-pawn, and eyeing the queenside, especially the b3-pawn, which we'll see in the game.>) 25.Qg2 (<? - A mistake, losing a pawn. Better was 25.Kh2 or 25.Kg2>) axb3 26.axb3 Qxb3 (<In hindsight, 26…Rxa1 followed by 27…Qxb3 might have been better, since white doesn't have the option of the c4-c5 push, which will give him some dangerous counterplay in the game. On the downside, white hs now control of the a-file.>) 27.Rxa8 Rxa8

click for larger view

(<Here, I felt really good about my position, I thought that at worst I have a draw, while white would have to defend passively in order to try and salvage something. A feeling of complacency was taking over me, when suddenly there was a rude awakening…>)

click for larger view

(<Another good move. Black has many ways to go wrong.>)

Qe3

(<Lady Luck must have smiled upon me. I was considering either 30…Qc3 or 30…Qe3, oblivious to the dangers of the impending 31.Bxf7 sacrifice, which I dismissed as desperate and "obviously" incorrect. I only chose 30…Qe3 because it's closer to the white king, in case I would be in need of a few checks, which is exactly why it's correct. 30…Qd3 would also be a good alternative. Let's see what happens after 30…Qc3:

click for larger view

31.Bxf7! Kh8 (Forced, we'll see why) and after 32.d7! b3 33.Be8 Bf8 34.f7 b2 35.Qf2 h5 36.Qf6 Kh7 37.Qxd8 Qxg3 with a draw by perpetual check. If 31…Kxf7, then 32.Qd5 Kf8 33.Qe6! and if 33…Re8, then 34.d7! and it's all over.>)

31.Bxf7+ Kxf7 (<I decided to accept the challenge, 31…Kh8 didn't seem to inviting.>) 32.Qd5+ Kf8 (<The only move.>) 33.Qb7 Kg8 (<Again, the only move, as are black's next two moves.>) 34.f7+ Kh8 35.Qe7 Rf8 36.d7 Qe4+ (<And here we can see the benefit of 30…Qe3 over 30…Qe3: after 37.Kh2 Qe2 (Also attacking white's rook, this wouldn't be possible possible after 30…Qc3) 38.Kg1 Be3 and black wins.>) 37.Kg1 Qd4+ 38.Kh2 b3 39.Rf6 b2 40.Rd6 Qf2+ 41.Kh3 b1=Q 42.d8=Q Qbf1+

(<The final position:

click for larger view

1.e4 e5 2.Nf3 Nc6 3.d4 (<I hate the Scotch! I really do. :-)>) exd4 4.Nxd4 Bc5 5.Be3 (<5.Nxc6 Qf6 6.Qd2 dxc6 is another common alternative.>) Qf6 6.c3 Nge7 7.Be2 d5 8.Nd2 (<According to the Opening Explorer this is a new move.>) Qg6

(<!? - In retrospect, 8...Bxd4 might indeed be the best move, as suggested by arsen387. After the proposed line 8..Bxd4 9.cxd4 Nxd4 10.Qa4+ Nc6 black is up a pawn, but white has some compensation for it. Rybka likes it best too, scoring it as +0.28 in black's favor at 18 ply, though it suggests 10.Bd3 instead of 10.Qa4.

8...Nxd4 9.cxd4 Bxd4? 10.Qa4 Nc6 11.exd5 is of course out of the question.

8...Bd7, simply developing, was another viable alternative.>)

9.Nxc6 (<9.Nb5!?>) Qxc6

(<Here I was calculating 9...Bxe3 too, but found it inadequate because of 10.Nxe7 Bxd2 11.Qxd2 Kxe7 12.Qxd5, where white is simply up a pawn and with black having no compensation.

I have missed, however, that I can play 10...Qxg2! first and after 11.Rf1 Bxd2 12.Qxd2 Kxe7 13.Qxd5 (13.exd5!?) Rd8 14.Qe5+ Kf8 it's more of a fight:

click for larger view

>)

10.Qb3 Qb6 11.Bxc5

(<?! - Lets me off the hook. After 11.Qb5! Qxb5 12.Bxb5 c6 13.Bxc5 cxb5 14.O-O-O white has much better prospects in the endgame than the one actually resulting in the game:

click for larger view

The following moves are mostly self-explanatory, but I will borrow some of arsen387's commentary anyway.>)

Qxc5 12.O-O O-O 13.exd5 Qxd5 14.Rad1 Qxb3 15.Nxb3

(<"The queens are exchanged, material is even, but my pieces are better developed" (arsen387).>)

(<"I like this move which accomplishes many things:

1) removes any back rank mate possibilities,
2) enhances my B's mobility
3) reduces black N's mobility, which now has very few squares to move" (arsen387)>)

Rfd8 21.Rxd8+ (<21.Rd1 Rxd7 22.Rxd7 Ne5.>) Rxd8 22.Re2 (<Protecting the 2nd rank.>) f6 23.Bg4 Ne5 24.Be6+ Kf8 25.f4 Ng6 26.b4

(<The final position:

click for larger view

"I think in final position I had a little advantage with my good bishop against the "bad" knight, but didn't know how to continue. Maybe pushing the doubled b-pawn could be good, but it could also lead to unwanted complications, so I offered a draw" (arsen387).

I think that white has all the chances here, that's why I was a little surprised by his draw offer, but it's also true that black has a solid position, albeit with less manoeuvring space. Defending passively isn't exactly my "forte", so I gladly accepted, since I didn't like my position very much. A good game, though it seems white was in complete control throughout and black never had a chance for more.>)

1/2-1/2

<ahmadov - TheAlchemist, Round 3>

1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Bg5 e6 7.Bd3 (<An unusual move, 7.f4 is almost automatic here, there are only 8 games in the Opening Explorer with the move played.>) Be7 8.O-O O-O 9.Qd2 (<According to the database, this is a new move. 9.Re1 was tried in Simon Terzi vs Sasikiran, 2005 and 9.h3 in I Gmirina vs O Shevchuk, 2003 >) Nc6 10.Nb3 b5 11.Rfe1 Bb7 12.Rad1 Rc8 13.Ne2 (<13.f4!? h6 14.Bxf6>) Ne5 (<Preparing the freeing d6-d5>) 14.Ng3 d5 15.exd5 Nxd3 (<The knight was hanging.>) 16.Qxd3 Nxd5 17.Bxe7 Qxe7

click for larger view

18.Qd4

(<This simply loses a pawn, though I haven't found any other alternatives that don't.

18.c3 Nf4 19.Qd7 Qxd7 20.Rxd7 Bxg2 black wins a pawn.

18.f3 Nb4 and a pawn must fall here too.

18.Qd2 Rfd8 19.Nd4 Nb4 20.c3 Nxa2 is a bit trickier, although black can save the knight after 21.Ra1 with 21...Bd5, but it seems really awkward. Better is 21...Qc5! (Rybka) and if 22.Rxa2? then the double attack 22...Qd5 wins the rook. If 22.f3 instead, then 22...e5. 21.Ngf5 Qf6 doesn't seem so good either. But I don't know if I would have gone into the Nb4-xa2 line, because the knight is temporarily out of play.>)

Rxc2 19.Rc1 Rfc8 20.Rxc2 Rxc2 21.h3 h6 22.Nf5 Qg5 23.h4

(<? - A blunder, but black was much better already. Rybka actually evaluates the best move 23.Ng3 as being 3.02 in black's favor at 16 ply, giving the continuation 23...Nf4 24.Re5 Ne2 25.Rxe2 Rxe2. At 17 ply it switches to 24.Qe3 Bxg2 25.Nd4 Rxb2 26.Ne2 Nxe2 27.Rxe2 Qxe3, but it seems hopeless for white, while something like 24.Qd7 would run into 24...Bxg2 and here too white's toast. I thought I had an advantage, but not by that much.>)

Qxf5

(<The final position:

click for larger view

I think I responded quite well to white's a little unorthodox approach, playing a typical "sicilian" setup with Nc6, b5, Bb7, Rc8 etc. (but that doesn't always work!). So, I guess that despite white's big blunder at the end, I should have been able to win anyway, though, as I said, it wasn't so clear to me at the time. Aside from the final blunder and the possible mistake at move 18, I guess white's choice of some opening moves wasn't so fortunate either, probably entering more usual lines would have been better.>)

<TheAlchemist - fictionist, Round 6>

1.Nf3 Nf6 2.g3 g6 3.Bg2 Bg7 4.O-O O-O 5.d3 d5 6.Nbd2 Bf5 (<According to the Opening Explorer this is a new move. 6...c5 seems the most natural here, and it's been the most successful for black. As fictionist himself mentioned, he was aiming at preventing e2-e4, the thematic break in the King's Indian Attack type of openings.>) 7.Nh4 Bg4 8.h3 Bd7 9.c4 c6 10.e4 (<White successfully gets in the e4 break.>) Qc8

click for larger view

11.g4

(<Somewhat more double-edged than the more prosaic 11.Kh2, certainly in my style. 11.e5 would have been a good alternative as well, which, of course, I didn't consider at all, since I was only focused on black's obvious threat of Bxh3.

After 11.Kh2 black would have many choices, but 11...e5 seems good, preventing the same move for white. 11...Na6, developing, would also have been logical.

After 11.e5 Nh5 Rybka recommends 12.Ndf3 Bxh3 13.Bxh3 Qxh3 14.Ng5 Qc8 15.g4 h6 16.Nxf7 Rxf7 17.Nxg6, but it's pretty close (+0.20 for white at 18 ply).>)

dxe4 12.dxe4 Rd8 (<I think this is a mistake. 12...h5 would have been preferable, and here the awkward 13.f3 seems best, since 13.g5 leads to 13...Nh7 14.Ndf3 Bxh3, while after 13.e5 black can play 13...hxg4! 14.exf6 Bxf6, regaining the piece.>) 13.e5 (<An important in-between move, maybe black missed it.>) Ne8 14.Qe2 Na6 15.f4 (<And here white's plan is pretty straightforward, though black is still fine.>) Rb8 (<15...f5 would have certainly been better, or even 15...Qc7.>) 16.f5

click for larger view

e6

(<Another mistake, maybe the decisive one. 16...gxf5 would have been preferable, leading to something like 17.gxf5 (17.Nxf5!?) b5 (Nb4!?), where white is better, but it's still probably not decisive. Though it's certainly not easy to play.>)

17.fxg6 fxg6 18.Qf2 (<The decisive move, black is toast. 18.Ne4 would also have been a good alternative.>) Qc7 19.Qf7+ Kh8 20.Ne4 Bc8 (<20...Qxe5 21.Nxg6!>) 21.Nxg6+ hxg6 22.Qxg6

(<And here black resigned. The final position:

click for larger view

>)

1.d4 Nf6 2.c4 e6 3.Nf3 b6 4.g3 Bb7 5.Bg2 Be7 6.O-O O-O 7.Nc3 Ne4 8.Bd2 (<The old variation, 8.Qc2 is an alternative, as is 8.Nxe4, which is very drawish, though.>) f5 (<8...Bf6 is slightly more popular.>) 9.d5 (<9.Qc2>) Bf6 10.Rc1 c5 (<10...Na6 was played more often: Opening Explorer >) 11.Ne1 Bxc3 12.Bxc3 exd5 13.cxd5 d6 (<Following A G Panchenko vs Balashov, 1992 (0-1).>) 14.Nd3 (<A new move, sacrificing a pawn.>) Bxd5 15.Nf4 Bf7

click for larger view

16.Nh5

(<!? - This is not a real sacrifice, since 16...Bxh5, which, incidentally, was played in the game, is followed by 17.Qd5, forking the king and rook.

The other possibility was 16...d5. Here, and interesting line would be 17.Bxg7!? Bxh5 18.Bxf8 Kxf8 (18...Qxf8? 19.Qd5) 19.f3 Nf6 20.e4, leading to the following position:

click for larger view

Here, black could try 20...Nbd7 (20...fxe4 21.g4) 21.exf5 d4 22.g4, which Rybka evaluates as +0.34 for black at 20 ply, suggesting 22...Bf7 23.g5 Nd5 24.f4 Ne3 25.Qd3 Nxf1 26.Bxa8.>)

Bxh5 17.Qd5+ Rf7 18.Qxa8 Bxe2 (<Now black has 2 pawns for the exchange, but he can't keep them.>) 19.Rfe1 Bd3 (<Or the immediate 19...Bb5, where white could play 20.Bxe4 or even 20.Rxe4 (20...Bc6 21.Re8!). 20.Qd5 Bc6 21.Bxe4 Bxd5 22.Bxd5 Kf8 23.Bxf7 Kxf7 wouldn't be so good.>)

20.Qd5 Bb5 21.Bxe4 fxe4

(<Here, 21...Bc6 wouldn't as good as in a note above, since after 22.Qxc6 Nxc6 23.Bxc6, we reach the following position:

click for larger view

Here, white would be much better. Rybka evaluates it as +1.81 in white's favor at 21 ply, suggesting 23...Re7 24.Rxe7 Qxe7 25.Re1 Qc7 26.Bd5 Kf8. White will then play Re6, Bd2-f4, while black doesn't really have any real counterplay.>)

click for larger view

33.b5 (<I thought this was a mistake, but it's actually a good move, as I will show later.>) d3 (<The logical reply, which I didn't really consider, or greatly underestimated.>) 34.Qg4 Bxb5

click for larger view

35.Bc3

(<? - An important inaccuracy, I could even dub it as a mistake, since it's throwing away a clear win. After 35.Re8! Bxe8 36.Rxe8 Rf8 37.Bc3 Qc7 38.Qe6 Qf7 39.Rxf8 Kxf8 40.Qc8 Qe8 41.Bxg7 Kf7 42.Qxb7, reaching the following position:

click for larger view

Here, 42...Qe7 is almost forced, because of 42...Kg8 (43...Kg6) 43.Be5!. After 43.Qxe7 Kxe7 white should win the ensuing ending without major problems.>)

g6 36.Re6 d2

click for larger view

37.Bxd2 (<Here, I decided to force a draw. I could have still played for a win with 37.Rxd6 dxe1 38.Bxe1 Nxd6.>) Qxd2 38.Rxg6+ hxg6 39.Qxg6+ Rg7 40.Re8+ Bxe8 41.Qxe8+

(<The perpetual check is unavoidable. The final position:

click for larger view

I think we both played creatively, we both made a few inaccuracies, so a draw is the most fair result I think.>)

1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Bg5 e6 7.f4 Be7 (<7...Qb6, 7...Nbd7, 7...b5!?>) 8.Qf3 Qc7 9.O-O-O Nbd7 10.Bd3 (<10.g4 is another alternative, with the idea of 10...b5 11.Bxf6 Nxf6 12.g5 Nd7 13.f5 with a dangerous attack.>) h6 (<10...b5!?>) 11.Bh4 (<11.Qh3!, but I am very stubborn sometimes. For example: A Planinc vs Najdorf, 1973 >) g5 12.fxg5 Ne5

click for larger view

13.gxf6 (<?! - So, here we are. 13.Qe2 is, of course, the more logical move: Opening Explorer . Anyway, although I wasn't exactly the first to play this specualtive sacrifice, I still regard it as somewhat "mine", since I discovered it independently. Both my tries so far have been losses, though:

1) 13...Nxf3 14.gxf3 Bf8 15.Rhg1 Bd7 16.f4 Qc5 17.Be2 b5 18.a3 Rb8 19.f5 b4 20.axb4 Qxb4 21.Nxe6 Bxe6 22.fxe6 fxe6 23.f7+ Kd7 24.Bxa6 Rh7 25.Rgf1 axb2+ 26.Kd2 Qb4 27.Bd3 Bg7 28.Bf6 Bxf6 29.Rxf6 Rf8 30.Rb1 Qd4 31.Rb7+ Kc8 32.Rc7+ Kb8 33.Rf1 Qe5 34.Nb5 Rg7 35.Rc6 Kb7 36.Rc7+ Kb6 0-1

2) 13...Nxf3 14.gxf3 Bf8 15.Rhg1 Bd7 16.f4 d5 17.e5 O-O-O 18.Be2 Kb8 19.Bh5 Bc8 20.Rd3 Rh7 21.a3 Bd7 22.f5 exf5 23.Nxd5 Qc5 24.Ne7 Ba4 25.Bd1 Bxe7 26.fxe7 Rc8 27.Bf6 Re8 28.Rg2 Rxe7 29.Bxe7 Qxe7 30.Nxf5 Qxe5 31.Rg8+ Ka7 32.Nd6 Bd7 33.Bf3 Rg7 34.Rxg7 Qxg7 35.Bxb7 Qg1+ 36.Rd1 Qxh2 0-1

I have posted both already at Sicilian, Najdorf, 7...Be7 Main line (B99).

A few more sample games:

S Azarov vs B Predojevic, 2006: 13...Nxf3 14.gxf3 Bf8 15.f4 Bd7 16.f5 O-O-O 17.Kb1 Kb8 18.fxe6 fxe6 19.Rhg1 Bc8 20.e5 dxe5 21.Nf3 Bc5 22.Rg7 Qb6 23.f7 Bb4 24.Nd2 Bxc3 25. Nc4 e4 26.Nxb6 Bxg7 27.Bxd8 Rxd8 1/2-1/2

Brenjo-Ardeleanu, Backa Palanka 2002: 13...Nxf3 14.gxf3 Bf8 15.Rhg1 Bd7 16.f4 Qc5 17.Nce2 O-O-O 18.Rg7 e5 19.Nf5 Bxf5 20.exf5 Qe3+ 21.Kb1 exf4 22.Bc4 Qh3 23.Bf2 Qxf5 24.Bd5 Kb8 25.Rxf7 1-0

Gashimov-Kunin, Moscow 1997: 13...Nxf3 14.Nxf3 Bf8 15.Rhe1 Bd7 16.Bg3 0-0-0 17.Bf1 Bc6 18.Rd4 Rd7 19.Red1 Rg8 20.Bf2 Rg6 21.Rc4 b5 22.Rcd4 Rxf6 23.R4d3 b4 24.e5 Rg6 25.Nb1 Bb5 26.Re3 Bxf1 27.Rxf1 Rxg2 28.Nh4 Rxh2 0-1

Robin01 also tried it successfully: 13...Nxf3 14.gxf3 Bf8 15.Rhg1 Bd7 16.f4 0-0-0 17.Kb1 Kb8 18.Rg7 Bc8 19.Be2 b5 20.a3 Rd7 21.f5 e5 22.Nd5 Qb7 23.Nb3 Qa7 24.Rg2 Bb7 25.Bf2 Bxd5 26.Bxa7+ Kxa7 27.Rxd5 Rc7 28.Bh5 1-0

It seems I'm the only one losing with it :-)

<I couldn't delay the capture any longer, since after e.g. 24.Rg1, trying to preserve the g7-pawn, black would counter it with 24…Re8 (not 24…Rd7 25.fxe6), and if 25.fxe6 Bxe6 26.Nxe6 Rxe6. This is what I saw and decided against it.

However, Rybka recommends going back, 25.Rd1 and then 25…b6 26.Nf3 exf5 27.Ng5 fxe4 28.Bxf7 e3 29.Re1 Qd7 30.Bxg8 Rxg8 31.Rxe3, which is scored as +1.11 in black's favor at 18 ply.

click for larger view

>) Qxd8 25.Bxf7 Rxg7 26.fxe6 Rxf7 (<Probably not the best move, but understandable. 26…Qf6 was a serious alternative, where black can win the pawn as well after Rxf7. After the text move, 27.exf7 Qf6 28. Nb3 Qxf7 29.Rxd6. Still, black has an advantage here, Rybka considers it to be +0.70 in black's favor at 20 ply, suggesting 27.exf7 Qf6 28.Nf5 Bxf5 29.exf5 Qxf7 30.Rxd6 Qxf5 31.Rd1 h3. After 28.Nb3 it suggests 28…h3 29.Ka1 Qxf7 30.Rxd6 Qf4 Rd2 (+0.91 for black at 19 ply).

If we now go back to 26…Qf6, after 27.Nf5 black needn't transpose with 27…Rxf7, but play 27…Rg2! instead. After 28.Nd5 Qg5 (28…Qe5!?) 29.a4 Rg1 30.Rxg1 Qxg1 31.Ka2 Bxe6! 32.Bxe6 Qxh2 black seems close to winning:

click for larger view

The final position:

click for larger view