Annotated Game #75: Colle System goes awry

This game occurred in the first round of the next tournament after Annotated Game #71.  I always enjoy facing the Colle System as Black, as it never really seems to go anywhere against my preferred Slav-type setup, which was played in this game.  I've observed that the Colle seems to work best against Queen's Gambit Declined type defenses, in which Black shuts in his light-squared bishop.

Here, Black exchanges off the light-square bishop immediately and then focuses on development as White goes pawn-hunting on the queenside.  White's sense of danger was not operating and after his queen is nearly trapped, he is forced to give back material in order to save it.  Although the material balance was then roughly equal (3 pawns for a piece), Black definitely had the superior position.

In the remainder of the game, Black passes up several active options for improving his position and pressuring White, which unfortunately has been a common characteristic of my games.  If I get nothing else out of these annotation efforts, they have certainly driven home the importance of playing actively with both pieces and pawns.  In this game, White also missed some active possibilities, including the remarkable 20. f4!? and the counterintuitive 26. b4, which loses the b-pawn but gains a strong positional advantage for White's passed pawns.  In both cases, the strategic idea would have been to effectively mobilize White's pawn majority, where he had a favorable imbalance (to use Silman's term).

White eventually goes for a draw by repetition after striking a tactical blow against Black's kingside and winning a pawn there.  My opponent evidently did not trust his own position due to Black's possible threats.  At the time I was perfectly happy to acquiesce, not seeing how I could make real progress against White, who was also higher-rated.  The final result seems justified in this case, given the board situation.  Had Black been looking to win, it would have been better tried earlier, for example with 18...Ne4!?

[Event "?"] [Site "?"] [Date "????.??.??"] [Round "?"] [White "Class B"] [Black "ChessAdmin"] [Result "1/2-1/2"] [ECO "D04"] [Annotator "ChessAdmin/Fritz/Houdini"] [PlyCount "58"] [EventDate "2006.??.??"] {D04: Colle System} 1. d4 d5 2. Nf3 Nf6 3. e3 Bf5 {a good development for the bishop regardless. In the Colle System, this leads to an exchange of the light-squared bishops, which helps reduce White's attacking chances.} 4. Bd3 ( 4. c4 c6 {would transpose into the Slav.}) 4... Bxd3 5. Qxd3 e6 6. O-O Bd6 { an excellent square for the bishop, although this move involves a pawn sacrifice.} (6... c6 {would be the solid alternative.}) (6... c5 {is more amibitious, but safe enough.} 7. Qb5+ Qd7) 7. Qb5+ Nbd7 8. Qxb7 O-O {Black has reasonable compensation for the material, behind ahead in development and having threats against White's queen.} 9. c3 $146 {this blocks a key diagonal for White, complicating the queen's safety as well as his own further development. There are no positive aspcts to the move that I can see and it was probably played by rote as a standard move in the Colle System.} c5 $11 10. Qc6 {White seems to want to use his queen to take on Black's army alone.} Qb8 { protects the Bd6 and cuts off the b-file from the White queen's potential use.} 11. Nbd2 a5 12. a4 {this further cuts White off, ensuring that Black will at minimum regain the lost material.} (12. Qa4 $11) 12... c4 $19 {futher tightenting the noose around the queen. Now there is no alternative to forcibly clearing the pawns away, at the cost of a piece.} 13. Nxc4 (13. Qb5 { may have been White's original intention, but the b3 square is no longer available for retreat.} Qc7 {then wins, with the intention of following up with ...Rfb8.}) 13... dxc4 14. Qxc4 {while material is roughly theoretically equal (three pawns for the knight), Black's much more active pieces and open lines give him a solid advantage.} Ng4 {Black intends to force a trade of the knight and then get the Nd7 into the game via f6.} (14... Re8 {is the plan that the engines prefer. Black could use the ...e5 pawn break to further activate his pieces. For example} 15. h3 e5) 15. h3 Nh2 16. Nxh2 Bxh2+ 17. Kh1 Nf6 (17... Bd6 {the immediate bishop retreat may be simplest.}) 18. Qe2 $17 ( 18. g3 Bxg3 19. fxg3 Qxg3 {would be fine for Black, given White's shattered kingside and looming threats after ...Ne4.}) (18. b4 $15 {is Houdini's preference, keeping things solid on the kingside while trying to generate counterplay on the queenside with the extra pawns.}) 18... Bd6 (18... Ne4 $5 $17 {would be the active way to play. Now g3 would be unplayable for White, as Black would end up taking back with the knight and a 3-way fork.}) (18... Bc7 { is a better retreat, avoiding the threat of a pawn fork on e5.}) 19. e4 $15 e5 {this plays into White's strategy, allowing his pawns to gain strength.} (19... Bf4 {would contain the newly-activated dark-square bishop.}) 20. d5 {obvious but not best.} (20. f4 $5 {is the attacking move the engines find. The f4 pawn cannot be captured due to the pawn fork on e5 and Black has to work hard to avoid major problems.} exd4 (20... exf4 $2 21. e5 {and now the Black pawn is not threatening anything, unlike after taking on d4.}) 21. e5 $11 Re8 {saves a crucial tempo by pinning the e-pawn.} 22. Qd3 dxc3 23. exf6 (23. exd6) 23... cxb2 24. Bxb2 Qxb2 25. Qxd6 Qxf6) 20... h6 {seeking to limit the scope of the Bf1 by taking away g5.} 21. Qf3 {threatening to win a pawn with Bxh6, which would leave the g7 pawn overloaded.} Nd7 {a passive retreat to an undefended square, which allows White's subsequent tactic.} (21... Be7 {would instead protect the knight and keep it on a more active square.}) 22. Bxh6 {now White can play this anyway, stripping Black's king of protection.} gxh6 (22... f5 $5 {is the defense the engines find, which is not at all obvious. The point being that after} 23. exf5 gxh6 24. Qg4+ {is no longer is a double attack on the Nd7. Black also now controls the f6-square and the king can escape via f7 if needed. }) 23. Qg4+ $11 {Theme: Double Attack, notes Fritz.} Kh8 24. Qxd7 Ra7 25. Qf5 Rb7 {this move in reality makes Black more vulnerable to a push of the b-pawn, rather than threatening it.} (25... f6 $5 26. b4 axb4 27. cxb4 Qxb4 {and Black can hold.}) 26. Rab1 {the b4 push probably doesn't occur to White either, as it would give back a pawn.} (26. b4 axb4 27. cxb4 Rxb4 $16 {White would possess a major advantage in the form of the two passed pawns and Black's weak king position, along with a very active queen.} 28. Qf6+ Kh7 29. Rfc1) 26... Rg8 27. Qf6+ Kh7 28. Qf5+ {White essentially offers a draw with this move.} Kh8 {and Black is happy to take it.} (28... Rg6 {would be a little stronger (and completely equal, according to Houdini), but I understood that the invited repetition would lead to an immediate draw.}) 29. Qf6+ Kh7 1/2-1/2

The importance of CCT: example 5 - London Chess Classic Round 9

As another entry in the ongoing series featuring the importance of CCT (Checks, Captures and Threats), here is the final round game from the recently-completed London Chess Classic between Hikaru Nakamura and Luke McShane.  At first glance, on move 32 Black appears fine, with his Ne5 protected and attacked twice after White's 32. R1d5 move. However, a CCT check would reveal that this is not the case and the "obvious recapture" after the move played (33. Qxe5+) simply does not work, due to the Bd7 which will be left hanging at the end of the sequence.

Annotated Game #74: Round 3 - Round Turkey Tournament

This final round of the 2012 Round Turkey Tournament was the decisive one, as any of the players could theoretically have won it.  I had no idea how Tim Clark (aka Moth) was going to open the game, although I did prep the Caro-Kann Advance variation to some extent, since it is a popular choice for White these days.

Black has a relatively easy time of it in the opening and by move 9 has his pieces comfortably placed.  I decided not to get too fancy in the early middlegame and was thinking about quietly increasing the pressure on the d-pawn when White threw in the tactical surprise of 15. Nxd5.  Black is objectively fine here, but the failure by both of us (apparently) to spot the key ...Nxe5 countermove - made possible by the unprotected White queen on d3 - is quite instructive.  I had the mental assumption when both queens were on the d-file that opening the file would not do anything, only seeing the ...Nxe5 possibility once I had a rook on d8.  This is a good example of how doing a general tactical status check can be a help (and should be a regular feature of one's thinking process).  (I first saw this idea expressed in Understanding Chess Tactics by Martin Weteschnik.)

Despite missing the best reply, I manage to hang on and after the sequence is completed, regain equality.  White nevertheless retained what initiative was left in the position and I soon felt under pressure again after he pushed in the center with 24. d5.  Further inaccurate defending by Black leaves him with a somewhat scary-looking position as of move 29, although it was still objectively OK.  Attempting to counterattack in the center, I play Rxe3, which would have lost had White taken the fleeting opportunity to play d6 that was presented by Black's overloaded pieces.  Luckily for me, I immediately extricated myself and then was able to head for a setup that would force perpetual check.

My opponent didn't want to accept a draw, though, so decided to roll the dice with a rook exchange that lost him two pawns, leaving us with a R+P endgame featuring three Black kingside pawns versus two White kingside pawns.  With time growing shorter, White got very aggressive and failed to do a CCT (checks, captures and threats) check on move 41, allowing White's rook to check and then pick up the b-pawn.  The end came quickly afterwards.

My thanks to Tim for playing an interesting and strong game, which gave me a lot to look at during analysis.

For anyone else who wants to join the fun during the next cycle, the 2012 Double My Egg Nog FICS tournament still has a space available.

[Event "rated standard match"] [Site "Free Internet Chess Server"] [Date "2012.11.24"] [Round "?"] [White "Timmmmm"] [Black "ChessAdmin"] [Result "0-1"] [ECO "B12"] [WhiteElo "1723"] [BlackElo "1547"] [Annotator "ChessAdmin/Houdini"] [PlyCount "92"] [EventDate "2012.??.??"] [TimeControl "3600+5"] {B12: Caro-Kann: Advance Variation} 1. e4 c6 2. d4 d5 3. e5 {the Advance Variation is now the de facto main line, in terms of frequency of play at the professional level.} c5 {Black is willing to sacrifice this pawn for improved piece play. White decides not to accept.} 4. c3 Nc6 5. Nf3 Bg4 {NOT ...e6 before letting the bishop out.} 6. Be2 e6 7. Be3 Nge7 {Qb6 is the major alternative in the database. I instead opt to get the kingside developed.} 8. h3 {leading to a trade of minor pieces, as retreating the bishop would cramp Black and invite the g4 thrust.} Bxf3 9. Bxf3 Nf5 {here I consider Black's position as very comfortable, with superior activity for his minor pieces.} 10. O-O {around here is when I start thinking about playing ...Qb6, but never do.} cxd4 {an amateur move, relieving pressure on d4 just to make things less complicated.} (10... Qb6 {is Houdini's choice, protecting c5 again and pressuring b2. At the time, I didn't like the fact that taking multiple times on d4 would just end up with a Black knight pinned to the queen by the Be3, but there were other good reasons for the queen development, to ratchet up the pressure on White.}) 11. cxd4 Bb4 {this seemed to me to have better long-term prospects than development to e7, and gives the bishop the retreat to b6 (to target d4) as an option.} 12. Nc3 O-O 13. Qd3 $146 {Whtie is now out of the database (with Houdini's preferred move). This connects the rooks and usefully centralizes the queen on the b1-h7 diagonal.} Rc8 {Black completes his initial development.} 14. Rac1 Ba5 {after thinking for a fair amount of time on my initial middlegame plan, I had settled on pressuring d4 some more, with ...Bb6 to follow.} 15. Nxd5 {this came as a tactical surprise, using the theme of the overloaded pawn at e6, which was protecting both f5 and d5.} (15. Bg4 {is Houdini's preferred alternative.} Nxe3 16. fxe3 Qg5 $11) 15... Nh4 {Black has a cramped position, comments Houdini. This move is certainly a little awkward.} (15... exd5 $2 16. Qxf5) (15... Nxe3 {would have been the easier route.} 16. Nxe3 Nxe5 {; however, I didn't spot the possibility of capturing the e5 pawn while exploiting the pin on the d-file until later.}) 16. Nc3 (16. Nf4 { was necessary to protect the hanging Qd3 and therefore prevent the threatened Nxe5; however, neither my opponent nor myself spotted the capture at this point, due to a failure to recognize the status of the queen.}) 16... Nxf3+ { this line still gives Black an advantage, even if not optimal.} (16... Nxe5 17. Qe4 Nexf3+ 18. gxf3 $17 {and White's position is full of holes.}) 17. gxf3 $15 Qh4 $6 {I had fixated on the idea of penetrating with the queen here, although I knew White could easily defend in the short term.} (17... Nxe5) 18. Kh2 $11 Rcd8 {Interestingly, it was only when considering this move, where the rook takes the queen's place on d8, that I spot the Nxe5 threat. The move increases pressure on the backward pawn d4 and re-establishes the pin.} 19. Ne2 {this reduces the effectiveness of the knight and still allows Black to regain the pawn.} (19. Qe4 {immediately would probably be the best way for White to resolve his issues via simplification.} Qxe4 20. Nxe4 Bb6 $11 {and now that Black's queen is no longer on the board, he can't take advantage of White's kingside weaknesses, although he will have easy equality after regaining the d4 pawn.}) 19... Nxe5 20. Qe4 Qxe4 21. fxe4 {I thought for some time here about the various knight jumps, but eventually settled on retreating to c6, not seeing what I would get concretely out of ... Nd3.} Nc6 (21... Nd3 22. Rb1 $11 {is still Houdini's preferred line. The point is that the Nd3 can't be directly challenged by White without creating other problems in his position.}) 22. Rg1 {a rather transparent threat to play Bh6 as a follow-up.} g6 {the most obvious way to defend.} (22... Rfe8 {is a better defense, removing the rook from the line of fire. This would have effectively saved a tempo for Black - g6 does nothing for him - as well as better anticipating White's coming d5 push.}) 23. Rgd1 Bb6 24. d5 Bxe3 25. fxe3 Ne5 {I had to think about this one for a while as well, although there was less of a choice available than with the previous knight move. Houdini approves.} 26. Rc7 {the critical (and obvious) try for White.} Rc8 27. Rdc1 {with time getting shorter, my thinking process started breaking down around here, as I didn't even consider this possibility for White.} (27. Rxb7 {fails due to} Rc2 {which will win the Ne2 because of the pin on the Kh2.}) 27... Rcd8 {the riskier play, as at the time I didn't like the alternative.} (27... Rxc7 {is fine, however as} 28. Rxc7 exd5 29. exd5 Rd8 30. e4 Rd7 {and Black is equal, being able to stop the pawns with his piece blockade.}) 28. Nf4 exd5 29. exd5 {Black is still OK, although that pawn on d5 sure looks good for White.} Rfe8 {continuing the aggressive and risky play} (29... b5 {is Houdini's inspired choice, creating an outpost for Black on c4 that allows him to compensate for White's rampaging rook on the 7th rank.} 30. Rxa7 Nc4) 30. Rxb7 $14 Nf3+ 31. Kg2 {White threatens to win material: Kg2xf3} Rxe3 $2 {this is shown by Houdini to lose for Black, although it's not obvious at this point why it should.} (31... Nh4+ 32. Kf2 Nf5 $14) 32. Kf2 $18 {one should not forget that the king can also fork (double attack) different pieces!} Rde8 33. Rcc7 (33. d6 $1 {instead would win for White. Now the Re8 is overloaded and Black's pieces cannot protect each other and at the same time stop the advancing d-pawn.}) 33... Ng5 $11 {now Black's pieces are no longer overloaded and the position is equal.} 34. h4 Rf3+ { I thought for some time here and concluded that Black should get a perpetual check out of this sequence.} 35. Kg2 Rxf4 36. hxg5 Rf5 {now the king would not be close enough to the rook to be able to fend off the checks by counterattacking.} 37. Re7 $2 {White desperately wants to win, so plays riskily to prevent the perpetual.} (37. d6 {would lead to the perpetual check scenario, as Black would have no alternative that could prevent the pawn from queening.} Rxg5+) 37... Rxe7 38. Rxe7 Rxg5+ 39. Kf3 Rxd5 40. Rxa7 {This was the position we both had seen on move 37, although I'd evaluate it as at least a sure draw for Black, with winning chances. If White's pawns were further advanced, perhaps it would be another story.} h5 41. a4 {White was getting a bit short on time here and evidently didn't do a CCT check.} (41. Ke3 $17) 41... Rd3+ $19 {the b-pawn is now doomed.} 42. Ke2 Rb3 43. a5 {attempting to run for glory. With the rook in front of it rather than behind it, though, the pawn can never reach the 8th rank alive.} Rxb2+ 44. Kf3 Ra2 {Black now has a completely won endgame.} 45. Kf4 Kg7 46. Kg5 {White's last-gasp attempt to stop the Black pawns.} Ra4 {The rook closes off the king's escape and looms ready for the mate. Timmmmm resigns} (46... Ra4 47. Rxf7+ Kxf7 48. Kh6 Rxa5 49. Kh7 h4 50. Kh6 Rh5#) 0-1

How Carlsen makes us feel better about chess

The Financial Times this weekend has another article on chess, this time in its Lunch with the FT column.  Each week a personality - perhaps the best way to describe the people interviewed - has lunch with one of the FT's reporters and has an informal interview and conversation.  This week, it's Magnus Carlsen, who gave the interview just before the start of the London Chess Classic.

Similar to Kramnik's commentary in "How Kramnik makes us feel better about chess", I found Carlsen's approach to chess and his views on playing and training to be refreshing, having a simplicity and mature clarity about them.  When top players matter-of-factly discuss how positions are unclear and admit their own limits, it's an important lesson for improving players as well.

The article is worth reading in its entirety, but I've excerpted some of what I consider the most relevant points on mental attitudes, the importance of developing intuition and the role of planning.

• "...what at first seems like studied indifference is a genuine character trait of not easily becoming worked up, of taking things in one’s stride rather than needing to feel always in control. In fact, Carlsen seems unfazed by many things, among them not knowing whom he is playing when, how well he has to do in the London Classic to beat Kasparov’s record, or, for that matter, where to meet for lunch." 
• "...Carlsen says the difficulty with being tired when playing chess is that things don’t come intuitively. I point out that the stereotypical image of the game is that it is won not through intuition but hyperrational analytical powers. 'Of course, analysis can sometimes give more accurate results than intuition but usually it’s just a lot of work. I normally do what my intuition tells me to do. Most of the time spent thinking is just to double-check.'"
• "Still intrigued by the claim that intuition has pride of place, I ask him about the importance of spontaneity in chess. 'Of course, you make plans but the positions are often too complicated for proper planning. Then suddenly you get an idea.'"

Carlsen also emphasized the importance of being able to feel the joy in chess, rather than having it become a grind.  This and all of the above points I think are quite applicable to the improving player.

Annotated Game #73: Round 2 - Round Turkey Tournament

Round 2 of the 2012 Round Turkey Tournament featured a struggle with Rocky Rook in the English.  The first three moves produced an Old Indian-style formation on Black's part, as was previously seen in Annotated Game #35.  Unlike with that game, where White played an early d4 to exchange in the center and then fianchetto his bishop on g2, this time a central strategy is followed, with a very different-looking game.

Rocky's decision to push 5...e4 I think determined the whole strategic character of the game.  I was reasonably familiar with the idea, having looked at it in other closed-type English positions, so it didn't bother me too much.  Here I thought it was going a little too far out on a limb for Black, since the pawn would be difficult to support properly.  That is in fact what occurred, as White in the early middlegame is able to pick up the pawn.

I was worried about some of Black's counterplay shortly afterward, for example if he had chosen to penetrate on the second rank with 19...Rc2.  By the time he decides to try for counterplay a bit later on the kingside, however, I was able to calculate - after some initial trepidation - that it would come to nothing.  Once Black's threat was dealt with, I was able to find active, attacking continuations that increased White's positional advantage and eventually led to a mate threat.  This is in contrast to earlier in the game, where I passed up several interesting, aggressive continuations (8. g4!? and 19. e4 stand out) in order to put safety first.

Thanks again to Rocky for an interesting game and of course for his good work organizing the tournament.  I plan on participating in the next one on FICS, the Double My Egg Nog tournament, which still has one space free for an interested player.

[Event "rated standard match"] [Site "Free Internet Chess Server"] [Date "2012.11.21"] [Round "?"] [White "ChessAdmin"] [Black "RockyRook"] [Result "1-0"] [ECO "A22"] [WhiteElo "1556"] [BlackElo "1682"] [Annotator "ChessAdmin/Houdini"] [PlyCount "59"] [EventDate "2012.??.??"] [TimeControl "3600+5"] {A22: English Opening: 1...e5 2 Nc3 Nf6} 1. c4 e5 2. Nc3 Nf6 3. Nf3 d6 { Black thought for a bit before playing this, so I figured he was out of his personal book.} 4. e3 {the obvious alternative for an English player is g3, but I wanted to stick with a central strategy, having looked at similar position-types before. Previously (in annotated game #35) I had played d4 immediately.} Be7 5. d4 e4 {the big strategic decision in the game. Now play will revolve around attacking Black's central pawns. White scores over 85 percent in the database after this, although the sample size is very small.} 6. Nd2 (6. Ng5 {is also viable.}) 6... Bf5 7. Be2 (7. Qc2 {would force the issue in the center immediately, although I didn't particularly like some of the implications. For example} Bg6 {to remove the hanging bishop as a potential tactical weakness} 8. Ndxe4 Nxe4 9. Nxe4 d5 10. cxd5 Bb4+ 11. Bd2 Bxd2+ 12. Kxd2 {the king is forced to recapture, because of the hanging Ne4 if Qxd2.}) 7... c6 $146 {Secures b5+d5, observes Houdini through the Fritz interface.} ( 7... O-O 8. Qc2 d5 9. cxd5 Nxd5 10. Ncxe4 Bg6 11. Qc4 f5 12. Nc5 Bxc5 13. Qxc5 Nc6 14. O-O Re8 15. Bf3 Bf7 16. a3 Nce7 17. Nc4 b6 18. Qb5 a6 19. Qb3 Rb8 20. Ne5 Be6 21. Qd3 b5 22. Bd2 {Werts,R-De Vries,B Weijers 2004 1-0}) (7... Bg6 { is Houdini's pick.} 8. b4 $11) 8. O-O {I decided on safety first.} (8. g4 { is something that I actually did consider, however, and is Houdini's preference.} Bg6 $14 9. g5 Ng8 10. Ndxe4 {and now} Bxg5 {doesn't restore equality} (10... h6 $5) 11. Nxg5 Qxg5 12. h4 Qe7 13. h5 $16) 8... d5 9. cxd5 ( 9. Qb3 {immediately does seem best here, attacking b7 and not releasing the tension in the center.} Qd7 $14) 9... cxd5 $11 {at the time, I felt that this exchange opened up the queenside to White's eventual advantage while allowing White to pressure the center.} 10. Qb3 b6 $6 {the main problem positionally with this move is that it loses control of c6.} (10... Nc6 11. f3 (11. Qxb7 { would see White pawn-hunting without sufficient support for his queen. For example} Nb4 12. Bb5+ Kf8 13. Nb3 Rb8 14. Qxa7 Ra8 {and now Black gets a perpetual on White's queen.}) 11... exf3 12. Bxf3 $11) 11. f3 $14 {I was focused on getting in this break, as without it White's pieces will be smothered. The idea of course is also to break up Black's central pawns and keep pressuring them.} (11. Bb5+ {I underestimated the strength of this alternative, not realizing that blocking with Bd7 wouldn't work. White simply wins a pawn by force.} Nbd7 $16 (11... Bd7 12. Nxd5 Nxd5 13. Qxd5) 12. Nxd5 O-O 13. Nxe7+ Qxe7) 11... O-O (11... exf3 {I thought would leave Black with an OK game, although I'd prefer White's position. In any case, it was necessary to avoid the loss of a pawn in the game continuation.} 12. Bb5+ Bd7 13. Nxf3 Bxb5 14. Qxb5+ Qd7 15. Ne5 Qxb5 16. Nxb5 $14) 12. fxe4 $16 {the problem for Black now is his hanging Bf5, so he has no choice but to recapture with it.} Bxe4 13. Ndxe4 dxe4 (13... Nxe4 {is worse due to} 14. Nxd5) 14. Qc2 {I felt that Black must have missed this when initially calculating the exchange sequence. It's a bit counterintuitive that the e4 pawn would have no possible additional defenders. Also, retreats (in this case b3-c2) are typically harder to see mentally.} Re8 $2 {despite the engine's question mark, during the game I thought that this was the best practical try for Black, setting up some possible tactics along the e-file.} (14... Qc8 $5 $16 {is the defense found by Houdini, which postpones the fate of the e-pawn due to the pin on the Nc3. The engine still rates White with a strong positional plus, however.}) 15. Nxe4 $18 Nbd7 (15... Nxe4 16. Qxe4 Bg5 {is a possibility that I looked at for some time before taking the e-pawn, since if White snatched the Ra8 it looked like it could lead to some trouble. However, the simple Qf3 retreat would also be fine, so I made the practical decision to go ahead with the pawn capture. Black chose a less tactical line, in any case.}) 16. Bf3 {I didn't think that Black would actually miss the Nxf6+ discovered attack on the Ra8, but thought the bishop would be better on the long diagonal anyway. This comfortably holds White's advantage. Houdini however points out the attacking value was better on c4.} (16. Ng5 $5 {would set up a potential sacrifice on f7.} Rf8 (16... h6 $2 17. Nxf7 Kxf7 18. Bc4+ Kf8 19. Qg6 {with mate coming.}) 17. Bc4 $18) 16... Rc8 $16 17. Nxf6+ Nxf6 18. Qf2 {this was the other reason for moving the bishop to f3, to clear the second rank for the queen move. I was not thinking creatively enough, however; the same benefits from moving Qf5 later would apply here, as well as removing the potential pin on the second rank after Bd2. } (18. Qf5 $16) 18... Rc7 {Houdini doesn't like this, for reasons which will shortly become apparent.} (18... Bd6 $5 19. g3 $14) 19. Bd2 {this is part of what should be a simple plan - connecting the rooks and dominating the c-file - but I had to look at the possible rook invasion on c2.} (19. e4 $5 $16 { I considered this, which would get the central pawn roller moving, but thought it would be too risky.}) 19... Bd6 (19... Rc2 20. Rfc1 $5 {was my primary line} (20. Qe1 {is better}) 20... Rxb2 21. Qe1 {with the idea that now Black has to watch out for his rook being trapped.}) 20. Rac1 $16 {now White can dominate the c-file, since Black's rook on c7 - while defended twice - can be exchanged off at will and Rc1 then played.} Ne4 {I had seen this possibility and decided that it wasn't worth trying to avoid. The exchange is essentially forced, but leaves White that much closer to a winning endgame. The trade also removes an obstacle to White pushing e4.} 21. Bxe4 Rxe4 22. Qf5 {I selected this because it was the most active option for the queen and posed several problems - the immediate attack on e4, the potential control of c8 in tandem with a rook (which is what eventually happens), and an attack on h7. Regarding the last possibility, White has some threats involving a rook lift (Rf1-f3-h3) and the Bd2 would be a great help in a kingside attack once e4 is played.} Rh4 $6 { I had looked at this variation and thought it was superficially attractive, but in the end lead nowhere for Black.} (22... Re8 23. Qf3 $16) 23. g3 { the obvious response, blocking the attack on h2 and kicking the rook.} Rh6 $2 { now if Black had the time to play Rf6, he could solve his problems. But he doesn't have that crucial tempo.} (23... g6 24. Qf6 Qxf6 25. Rxf6 Rxc1+ 26. Bxc1 Be7 $16) 24. Rxc7 $18 Qxc7 25. Rc1 Qb7 {loses by force, although even after something like . ..g6 Black has a hopeless endgame.} (25... g6 {there is nothing better in the position, says Houdini.} 26. Rxc7 gxf5 27. Rxa7 Bf8 $18) 26. Rc8+ Bf8 27. Bb4 {I have to admit that I stopped calculating at this point, seeing no viable way out for Black.} g5 {I told Rocky in the post-game discussion that this was a fiendishly clever try. It certainly made me work for the win. the point is that the rook can now contribute defensively, unlike after ...g6.} 28. Bxf8 {not the quickest way, but still a mate.} Rc6 {stopping Bh6 delivering mate.} 29. Qxg5+ {it took me a while to figure out an accurate continuation. Among other things I considered bailing out into a won queen endgame by exchanging rooks on c6, which however would have been a real struggle. The necessary follow-on Qd8 was not obvious to me, at first.} Rg6 { forced. Now White finds the move that protects the rook and keeps the back rank mate threats alive. Black will have to either give up the queen or accept mate immediately.} 30. Qd8 {RockyRook resigns} 1-0