Here's my computer-checked analysis:

<31. Bh6+!! Kg8>

Black declines to take the Bishop, since White wins decisive material due to the mate threats created after 31... Kxh6 32. f6! From here, play could continue 32...g5 (32...Kg5 33. h4+! Kxg4 (33... Kh6 34. Qd2+ g5 35. Qxg5#) 34. Kh1 Qc3 35. Rg1+ Kh5 36. Qg2 ) 33. f7 Qe7 (33... Rf8 34. Qf6#) (33... Kg6 34. fxe8=Q+ Kg7 35. Qff7+ Kh6 36. Qf6#) 34. fxe8=Q Qxe8 35. Qf6+ Qg6 36. Qf8+ Qg7 37. Qxa8 .

<32. fxg6!>

This creates a new set of mate threats, which can only be temporarily avoided and at the expense of decisive material loss.

<32...Nd8>

If 32... hxg6, then White is winning after 33. Qf6 d3+ (33... Qe7 34. Qxg6+ Kh8 35. Bg5 Rg8 36. Bf6+ Qxf6 37. Qxf6+ Rg7 38. Rb3 ) 34. Ne3 Qe7 35. Qxg6+ Kh8 36. Bg5 Rg8 37. Bf6+ Qxf6 38. Qxf6+ .

If 32... Qe7, then White is winning after 33. Qf5 Qe6 34. gxh7+ Kh8 35. Qg5 Qe7 36. Qg6 Nd8 37. Bg5 Bxe4 38. Bf6+ Qxf6 39. Qxf6+ Kxh7 40. Rb3 .

<33. Qf6 Qc7 34. Rf1 1-0>

Black resigns in the face of White's multiple, decisive mate threats. For example if 34...hxg6, White scores the full point with 35. Qxg6+ Kh8 36. Qxe8+ Kh7 37. Rf8 Nf7 38. Rxf7+ Qxf7 39. Qxf7+ Kxh6 40. h4 Bxe4 41. g5#.

Material: Even with Bs of opposite color. The White pieces are more active than the Black, a difference that Bs of opposite can magnify. The Black K-side has a weak dark-square complex that the White Bc1 can exploit. The Pf5 can advance to control dark squares or capture to weaken light squares and open the f-file. The White Qf2 also can reach the dark squares, although it has defensive duties: the Black Qc5 and Pd5 are on the same diagonal as Qf2 and Kg1, and Qf2 defends Nc2.

Candidates (31.): Bh6+

31.Bh6+

Black can accept the sacrifice:

31…Kxh6 32.f6 (blocking the Kh3’s retreat and threatening 33.Qh4#)

Black has 2 feasible defenses.

(1) 32…g5 33.f7 (threatening 34.Qf6#, 34.fxe8=Q, 34.f8=Q+)

White gains at least Q for B.

(2) 32…Kg5 33.h4+ Kxg4

[33…Kh3 34.Qd2+ g5 35.Qxg5#]

34.Kh2 (threatening 35.Qg3+ Kh5 36.Qg5+)

Black must sacrifice his Q to stop mate, because of

34…h6 35.Rg1+ Kh4 36.Qe2+ Kxh4 37.Qg4#

Black can decline the sacrifice by moving to 1 of 4 flight squares:

(1) 31…Kf6 [or Kf7] 32.fxg6+ Ke6 [Kxg6 or Kg8 yield a fast mate]

33.Qf7+ Kd6 34.Qxe8 (threatening 35.Bf8+)

wins because of the resource 34…d3+ 35.Ne3 (which also influences the next line).

(2) 31…Kg8 [Kh8 is worse] 32.fxg6 (threatening 33.Qf7+ or 33.Qf6 then 34.Qg7#)

32…hxg6 [else, drop a P] 33.Qf6 Qe7 34.Qxg6+ Kh8

35.Bg5 (threatening 36.Bxe7 and 36.Bf6+)

Then I noticed that 31.Bh6+ was safe. That is, black cannot play 31...Kxh6 since 32.f6 traps the king on the h-file where the queen and g-pawn produce mate threats from which black cannot seem to escape.

This observation alone practically makes the move worth playing. Even if black recognizes the threat, I've gotten my bishop into the attack in a forcing manner, and it prevents ...Rf8 (black's main defensive resource).

As soon as I started checking the answer I almost hit myself for missing fxg6.

This observation alone practically makes the move worth playing."

General stuff first - there is indeed little better than practice, practice, practice. Take any opportunity to try to work out a puzzle. Look for alternative lines than those which were played. Follow the excellent lines suggested by the regulars here. Try to work out where the losing player went wrong - or why he resigned when he did. If someone asks a question or suggests a new line, try to work out if you can give an answer or improve on their analysis.

Over time you will build up your caculating ability and pattern memory. I think that a daily dose of Chessgames.com is better for you than playing lots of bad chess or buying too many books and chess programs. I should know - I have done too much of both in my time.

One thing I have benefitted from is trying to visualise the lines without moving the pieces. So when the likes of Dzechiel and Patzer2 suggest an interesting variation that I had not thought of, I try to follow it from the starting position only. That helps to build combinational strength.

A bit of specific advice on this particular position. The superstars on the site will sometimes analyse long lines for several variations. It is very instructive to follow their line of reasoning, but you should not feel intimidated if you cannot do something similar.

What you need to do is to find a line of play that you would feel confident playing. In this case, there is nothing wrong with playing Bh6+ if you have not analysed every black retreat. White has not sacrificed material, is not under direct threats from black and has improved his position. So once you have worked out that black cannot capture the bishop you can play Bh6+ with confidence.

31. Bh6+ I saw immediately, a <DECOY> to drag the king out of its shell. Since declining loses lamely as in the game, the only two lines you need to look at are:

I) 31... Kxh6 32. f6 Kg5 33. h4+ Kxg4 and here I saw 34. Rb3 was winning and stopped analyzing, though in a rated tournament game I would have taken the time to find that Kh1 or h2 mates even faster.

II) 31... Kxh6 32. f6 g5 33. f7 with decisive threats on e8 and f6.

If you did not even consider 31.Bh6+ you need to work on your tactical awareness- the Black king's lack of defenders should jump out at you as a chance to compel it forward and mate.

click for larger view

Rybka2.3.2a.mp.x64: <24-ply>

1. (0.62): 30...gxf5 31.gxf5 Re8 32.Ne1 Rg8 33.Qh4 Nb8 34.Bg5 Nd7 35.Kf2 Qd6 36.Nd3 Kh8 37.Be7

2. (0.72): 30...Rd6 31.Bg5 gxf5 32.gxf5 h6 33.Bd2 Qc4 34.Ne1 Kh7 35.Rc1 Qg8+ 36.Qg3 Qxg3+ 37.hxg3