Things fall apart

Pilfered as it was by the chemist from Brantham, and more productively appropriated by the great Nigerian novelist the Nobel committee studiously ignored, Yeats’ message in a bottle pitches up very nicely on the shores of the late middle game. Very often, the player who has dominated the middle game finds that, once he cashes his positional advantage in, the position explodes in his hands. He is still winning, but the play becomes randomised, time short, and the player is beleaguered by a sense of injustice that random tactics should be allowed to disrupt his strategy. I’ve lost plenty of games this way. I’ve read that getting an advantage is a different skill from keeping it, but they are both composites of many skills; the ability to keep control in the late middle game is completely different from endgame technique.

Two recent games for Pimlico and Dulwich illustrate this. In this first game, I was careful but still almost slipped; in the second, I was careless and lucky.

You need to activate javascript to enhance chess game and diagram visualization.
{In the first game, I am playing White with 75 minutes for the game and a 15 second increment. I learned afterwards that my opponent&#8217;s experience was entirely in rapid and blitz, and he still had forty-six minutes left at the end of the game.} 1. c4 c6 2. g3 d5 3. Bg2 Nf6 4. Nf3 e6 5. b3 Be7 (5... Bd6 {feels more normal to me in this semi-Slav structure}) 6. Bb2 h6 {My opponent was worried about Ng5 in some lines but it looks like a waste of a tempo.} 7. O-O Nbd7 8. e3 O-O 9. d3 a5 10. Qe2 Ne8 (10... Qc7 {may be better, as my opponent thought}) 11. e4 {If White allows Bf6 and the exchange of bishops his attacking chances are reduced.} 11... a4 12. Nbd2 Bf6 13. e5 {I felt this should give me an advantage.} 13... Be7 14. cxd5 cxd5 ({After} 14... exd5 {I was thinking of keeping the position closed and gradually pushing the majority though I think my opponent and the computer prefer e6.}) 15. Nd4 { <span class="PgnWidget-anchor-diagram">[]</span> In this French-like structure this knight is very strong, supporting an f5 thrust while impeding Black&#8217;s queenside development.} 15... Qa5 {Black is angling for Nxe5 Qxe5 Qxd2, an idea I simply missed.} 16. a3 Qa7 {We both felt Black went wrong here.} (16... axb3 {looks better when Black gets some play on the queenside.}) (16... Nxe5 17. b4 Qc7 18. Nxe6 {is the computer&#8217;s preferred option for Black.}) 17. b4 g6 18. Kh1 Ng7 19. f4 Kh7 20. Rac1 Nb8 21. N2f3 ({I spent twelve minutes trying to make} 21. Nb5 Qb6 22. Nc7 Ra7 {work but couldn&#8217;t find anything clear enough e.g.} 23. Nf3 Nc6 {and felt my position was good enough not to need dubious complications. But I was now more than twenty minutes down on time.}) 21... Bd7 22. g4 {White wants to play f5 but it&#8217;s not so easy.} 22... h5 23. h3 { <span class="PgnWidget-anchor-diagram">[]</span> } 23... hxg4 {Probably a serious mistake. Black&#8217;s pieces are well positioned to deter f5 but not to contest the h file, especially the g7 knight. Black&#8217;s position was already difficult but it is now probably lost} 24. hxg4 Rh8 25. Kh2 {An important move, making it possible to contest the h file.} 25... Nc6 ({Here my opponent thought for five minutes, considering} 25... g5 {after which the computer thinks} 26. Nxg5+ {is winning though I was trying to see whether I could make f5 work.}) 26. Kg3 Qb8 {Now I thought for six minutes, and couldn&#8217;t see a good defence for Black against my h-file attack. The computer confirms this.} 27. Rh1+ Kg8 28. Rxh8+ Kxh8 29. Rh1+ Kg8 30. Nxc6 {The clearing of the diagonal deters f6 and makes it harder for the king to escape.} 30... bxc6 31. Bf1 {The point of my play, getting the queen to the h file as soon as possible.} 31... c5 32. Qh2 Nh5+ {Giving a knight to slow the attack .} 33. gxh5 g5 (33... cxb4 {runs into} 34. hxg6) 34. bxc5 {White is winning and I had more than twenty minutes, but the king&#8217;s position and the opening of various lines justify care.} 34... gxf4+ 35. Kxf4 Qf8 36. Rg1+ Kh8 37. h6 {Six minutes thought on this, which blocks the h6-c1 diagonal at the cost of tying the queen down.} 37... Bxc5 38. d4 f6 {Chucking another bishop on the fire, a reasonable approach in a desperate situation.} (38... Bxa3 {can be met by} 39. Bxa3 Qxa3 40. Qg2 ({or} 40. Qh4 Qc1+ 41. Kg3 Rg8+ 42. Kh2)) 39. dxc5 fxe5+ 40. Ke3 ({My opponent expected} 40. Kxe5 {which is also fine but looks a bit hairy. I had planned Ke3 but realised at this stage that I would allow Qxc5. Fortunately I saw a response}) 40... Qxc5+ ({My original analysis had been} 40... d4+ 41. Kd2) { <span class="PgnWidget-anchor-diagram">[]</span> } 41. Nd4 {Not the only winning move, but a pleasingly emphatic one.} 41... Qc7 ({Of course} 41... exd4+ {losses the queen to} 42. Bxd4+) 42. Qg3 {Here my opponent resigned. I thought there might still be some fight and we looked at some amusing variations afterwards, but after Bc6 or Bb5 White can simply exchange queens with Qg7+ and then pocket the bishop, and the computer finds a mate on four with h7.} *
() -
Commented by

The second game was played two days later. I started cautiously enough, but relaxed at a crucial moment.

You need to activate javascript to enhance chess game and diagram visualization.
1. d4 {I am playing Black with 75 minutes for 30 moves and twenty for the rest of the game (a big improvement on the previous 15-minute finish). I had played my opponent several years before, when he must have been about fifteen; he&#8217;s improved since.} 1... Nf6 2. Nf3 c5 3. d5 (3. dxc5 {is a reasonable alternative when} 3... g6 {is said to be risky because the pawn is hard to win back.}) 3... d6 ({I think my official repertoire is} 3... e6 {heading for a Blumenfeld after} 4. c4 b5 {but I forgot this and was trying for a Benko.}) (3... b5 {is another way to head for something like a Benko}) 4. Nc3 (4. c4 b5 {could lead to a mainline Benko}) 4... e6 5. e4 exd5 6. exd5 (6. Nxd5 {does very well in a small sample of games and may be an objection to Black&#8217;s play}) 6... Be7 7. Bf4 O-O 8. Bd3 {My opponent thought this was wrong, and I&#8217;m inclined to agree. White can ask for positional play with a4 and Nc4, or can cattle queenside and attack; my opponent had the second in mind, but this doesn&#8217;t fit.} (8. Qd2 {followed by queenside castling}) (8. Be2) 8... Na6 9. O-O Nc7 (9... Nb4 {seems not to achieve much e.g.} 10. Be2 Bf5 11. Rc1) 10. Re1 b6 {This steady approach seems to suit the position.} 11. Qe2 Re8 12. Rad1 Bb7 13. Bc4 {I was glad to have pushed his bishop here, but mine on b7 isn&#8217;t doing much either.} 13... Qd7 14. a4 a6 15. h3 Bf8 16. Qd3 {Continuing to restrain b5.} 16... g6 { <span class="PgnWidget-anchor-diagram">[]</span> The position is delicately balanced, but now my opponent is afflicted by an idea.} 17. Ng5 Bg7 18. Nge4 {Attacking d6, but wrong.} (18. Ne6 {was suggested by the opposing top board but doesn&#8217;t work after} 18... fxe6 19. dxe6 Qc6) 18... Nxe4 19. Nxe4 b5 {I think my opponent had forgotten about this.} 20. Ba2 c4 21. Qg3 (21. Qa3 {is also possible.}) 21... Nxd5 22. Bxd6 ({After} 22. Nxd6 {the computer likes Black after} 22... Rxe1+ 23. Rxe1 Nxf4 24. Qxf4 Qc7 ({rather than my thought} 24... Bxb2 {which runs into} 25. Rb1 Ba3 26. Nxb7 Qxb7 27. axb5 axb5 28. Bxc4)) 22... Qc6 23. c3 {This looked sensible, but now I spotted a tactic.} (23. Nc5 Bxb2 {is good for Black.}) 23... Rxe4 24. Rxe4 Nxc3 25. bxc3 Qxe4 { <span class="PgnWidget-anchor-diagram">[]</span> Black is a pawn up with well placed pieces, and quarter of an hour ahead on the clock, while White&#8217;s light-squared bishop is incarcerated. Unfortunately, this is where the fun starts.} 26. Bb4 Qe2 {Closing in for the kill.} (26... Be5 27. Qg4 {and Black has many good options.} ({I seem to remember expecting} 27. Qe3 {but this is not such a good idea})) 27. Rd7 {Short of time, my opponent spots his best chance.} 27... Bc6 ({At some point I had intended} 27... Qe1+ 28. Kh2 Be5 {but White has} 29. f4) (27... Be5 28. f4) 28. Re7 (28. Rc7 {loses to} 28... Qe1+ 29. Kh2 Be5 30. f4 Qxg3+ 31. Kxg3 Bxc7) 28... Qxa2 29. Qc7 Bd5 {Still thinking I had everything covered.} ({Black can bale out here and keep a winning material advantage with} 29... Be8 30. Qb7 Qb1+ 31. Kh2 Rd8 32. Ba5 Qd3) 30. Qd7 Qb1+ 31. Kh2 {I thought he might resign at the time control. Then I realised I had completely lost control of the position.} 31... Qf5 32. Re8+ {I had missed this when I took the bishop but thought my defence still worked.} 32... Bf8 33. Qe7 {I had missed this too.} 33... Rxe8 34. Qxe8 { <span class="PgnWidget-anchor-diagram">[]</span> } 34... Qf4+ ({Not}34... Kg7 35. Bxf8+ Kf6 36. Qe7#) 35. Kg1 Qh6 ({Black can take a perpetual with} 35... Qc1+ {but I was relieved to find something better.}) 36. axb5 axb5 37. f4 {Trying to restrict Black&#8217;s queen. But in the queen ending, this pawn turns out to be vulnerable.} 37... Qg7 38. Kf2 Bc6 ({The opposing top board pointed out} 38... f6 {which should allow Black to unravel without giving up a piece; if the light-squared bishop is attacked it can eventually run to d3 as in the next note.}) 39. Qc8 Bb7 40. Qb8 f6 {Black can win like this by giving the piece back, but it&#8217;s not necessary.} (40... Be4 41. Qe8 Bd3 {followed by f6, Qf7 and Kg7 would be much simpler. For some reason I thought my light-squared bishop needed to defend the king- perhaps a correction from my earlier carelessness.}) 41. Bxf8 Qxf8 42. Qxb7 {Queen endings are distinctive in two ways; advanced passed pawns are particularly dangerous because the defending piece cannot blockade the pawn without help from the king: and king safety is important, as the defender plays for perpetual check or occasionally mate and the attacker tries to use a fork to get the queens off.} 42... Qd6 (42... Qc5+ {looks natural but is less sharp.}) 43. Qxb5 (43. Kf3 {might be better}) 43... Qxf4+ 44. Kg1 Qe3+ 45. Kf1 Qc1+ 46. Kf2 Qxc3 {Being two pawns up allows a pawn sacrifice to force a queen exchange and leaves the opposing king more exposed than my own.} 47. Qe8+ Kg7 48. Qe7+ Kh6 49. Qf8+ Kg5 50. Qc5+ Qe5 {Using the threat of a queen exchange to protect the c pawn.} 51. h4+ (51. Qxc4 Qf4+ 52. Qxf4+ Kxf4 {and the pawn ending is winning e.g.} 53. g3+ Ke4 54. Ke2 f5 55. Kf2 Kd3 56. Kf3 g5 57. h4 h6 58. Kf2 Kd2 59. Kf3 Ke1 60. Ke3 Kf1 61. Kf3 Kg1) 51... Kg4 ({not} 51... Kxh4 52. Qxc4+) 52. Qc6 {Reginam expellas furca, tamen usque recurret.} (52. Qxc4+ Qf4+) 52... f5 { <span class="PgnWidget-anchor-diagram">[]</span> I played this and then thought I was getting mated.} 53. Qf3+ Kxh4 54. g3+ Kg5 (54... Kh3 55. Qh1+ Kg4 56. Qh4# {was the line that had scared me. But the h pawn&#8217;s removal has given the king freedom.}) 55. Qh1 {A last mate threat.} 55... Qd4+ 56. Ke2 Qe4+ {and my opponent resigned.} *
() -
Commented by