*****

<whiteshark: Learn from the mistakes of others. You won't live long enough to make all of them yourself> priceless!

Quite right. And the trick of turning a queen upside down to represent a rook has never caught on. Which is a shame.

This one took me a bit longer than usual because I was determined to make 33.Bg7+ work. Fritz confirms that it does work, but slowly. Here's one line:

33. Bg7+ Kg8 34. Bxh7+ Nxh7 35. Re8+ Rf8 36. Bxf8+ Ng5 37. Qxg5+ Kf7 38. Qh5+ Kxf6 39. Rf1#

In human mode I didn't quite get that far. After a while it dawned on me that this puzzle was all about pieces that you don't want to move, and men wearing make-up.

I didn't want to move my Re7 because that was blocking the black rooks from connecting with each other and defending along the seventh rank.

I didn't want to move my Bh6 because it was doing a great job of restraining the black king and building a mating net.

So while it was tempting to play Bg7+ and set up a discovered check, I couldn't do much with that bishop after the check other than put it on the h6 square that it was standing on already. 33. Bg7+ Kg8 34. Bxf8+ Kxf8 looked messy. I'm giving away a great attacking piece for a meh defender. White might be winning, but I'm not feeling the love for the position any more.

So if I don't want to move rook or bishop, what is left? The principled move would be to bring the Ra1 into the attack with something like 33.Rae1, but that doesn't feel like a puzzly move.

By a process of elimination, we arrive at a queen move. Mate looks to be near, so we have permission from Blessed Caissa to be outrageous. Looking across the board, what is the glam rock move complete with high heels, sequinned jumpsuit and KISS style demon make-up?

Ah yes, 33. Qg7+. The rook takes, pawn takes, pawn takes. It's a check-check-checkmate. And suddenly we have jumped genres from KISS to David Bowie's ch-ch-changes.

Definitely an easy Tuesday, white abuses the f8-h6 diagonal with a queen sacrifice (make up from yesterday?): <33.Qg7+ Rxg7 34.fxg7+ Kg8 35.gxf8=Q# 1-0>. That said, 33.Bg7+ is probably mate too and 33.Bxf8 should win as well, black's kingside is just busted too much. There are some variations to go wrong, whatsoever.

<Phony Benoni: Promoting to a queen is definitely best here. Your flag might fall before you found a third rook.>

Pattern-matching made me eliminate the queen move first! ... but I failed, and never reached anything else.

The laws of chess say that a move is completed when:

"in the case of the promotion of a pawn, when the pawn has been removed from the chessboard and the player's hand has released the new piece after placing it on the promotion square."

Checkmate immediately ends the game - you don't have to press the clock after delivering the mating move.

If I am right, the sequence is ...

1. Push your pawn to the queening square.
2. Replace the pawn with your chosen piece.
3. Take your hand off the new piece.
4. You win as long as you have some time left on your clock. It does not matter if your flag falls after you have completed the mating move (ie after you have removed your paw from the piece).

If your flag falls before you complete the move, your opponent can claim a win on time (or a draw if there isn't enough material) ... and this when you hope that the arbiter is standing nearby to spot exactly when the flag falls.

If I am right, it is the same with most draws (insufficient material, stalemate, etc). The game is drawn the instant that the move is completed (ie taking your hand off the piece) and not when you press the clock.

In trying to find an improvement for Black, the most interesting aspect of the game for me was the opening.

In particular, the move 8...a5 (which first appears in the OE database with Black's win in Leko vs Lputian, 1995) is one I find both interesting and puzzling. The computer doesn't find 8...a5 to be bad (i.e. equal evaluations), but it's one of the least popular choices among Masters.

According to the chessgames.com opening explorer data base, there have been 175 games played with the move 8. f4.

Of those 175 games, the reply 8...a5 has been played in only six of those 175 games (3.4%). Much more popular than moving this wing pawn are moves that develop a piece:

8...Bd7, as in Black's win in A Gabrielian vs R Khusnutdinov, 2013, was played in 81 of the 175 games after 8. f4 (46%), with the result being 33% wins for White to 31% wins for Black.

8...Ne7, as in Black's win in Schmittdiel vs A Berelowitsch, 2012, was played in 60 of the 175 games after 8. f4 (34%), with the result being 43% wins for white to 23% wins for Black.

8...Nc6, as in the draw in T Baron vs C Bauer, 2014, was played in 12 of the 175 games after 8. f4 (7%), with the result being six wins for White and one win for Black.

8...Nh6, as in Black's win in Z Almasi vs Hug, 1994, was also played in 12 of the 175 games after 8,. f4 (7%), with the result being four wins for White and three wins for Black.

8...a5, as played in this game, appears in six games in the opening explorer at http://www.chessgames.com/perl/ches..., with the result being four wins for White and two wins for Black.

While it's difficult to fault 8...a5 with concrete analysis, my own preference would be to go for one of the more frequently played piece developing moves (i.e. 8...Bd7, 8...Ne7, 8...Nc7 or 8...Nh6) over 8...a5.

However, Deep Fritz 14 finds 8...a5 to be just fine, and doesn't find fault with Black's play until <15...f6> is played.

Instead Deep Fritz 14 recommends 15...Rc8 (+0.36 @ 27 depth) as an improvement:

15...Rc8 16.Qf2 Be8 17.Bxf5 exf5 18.0-0 g6 19.Nxd4 Nd7 20.Be3 Qc3 21.Rfd1 Nc5 22.Ne2 Qc4 23.Bxc5 Rxc5 24.Nd4 Qb4 25.Rdb1 Qc3 26.Rd1 Rc4 27.Nb3 Qxc2 28.Qxc2 Rxc2 (+ 0.36 @ 27 depth)

The immediate 33. Bg7+ feels really good because (a) it's a forcing move and (b) we compel Black to walk into the cross hairs of a bishop-queen gun. Clearly 33...Rxg7 is not much of a defence.

This means that 33.Bg7+ feels solutiony. And if that isn't a word yet it darned well ought to be. It's a forcing move, it has a defence that doesn't work, it sets Black up for a bishop-queen battery. If it looks like a duck, quacks like a duck and is served with plum sauce and not enough pancakes then it's a ...

Oh no, it isn't. 33.Qg7+ is better but it's a harder move to spot because it does something unusual. It gives up a queen in order to create a pawn fork. Ordinarily, queen sacrifices don't lead to pawn forks - because what could a pawn fork that would be worth more than a queen?

What makes this particular pawn fork work is that the Black king is already mostly in a mating net and the pawn fork leads to a queening/ rooking promotion which gives mate.

The white rook that black has covered and threatens to capture screens the zig-zagging f-pawn for the one move it has to exist on g7 before it captures the knight to give mate on promotion.

White to play 23.

click for larger view