R Vera vs J Becerra-Rivero, 1996 (37.?)

White to play and win.

Material: N+P for B. The Black Kg8 is stalemated, and White has a forcing sequence.

Candidates (37.): Nh5

37.Nh5 (threatening 38.Qg7#)

There are exactly two troops:

Troop #1 is a knight that happens to be ready to attack the mating g7 square via <37.Nh5>. It can (and must) be captured of course by <37...gxh5>, but this opens the g-file.

Troop #2 is a rook that happens to be ready to attack an open file, such as the newly opened g-file, again with mate in view: <38.Rg1#>.

<37.?>

Target: 1:25;000
Taken: 0:31;150

Material: White up, + vs

Candidates: Qg7†, <[Nh5]>

-ML-
It's rather surprising how White can be up a pawn here. There is no compensation for Black whatsoever, and the kill from White is imminent. First I looked at the most aggressive move otb, Qg7†, when I spotted the key move.

<37.Nh5> and Black can only postpone mate on g7. The capture:

I certainly agree that 31. Nf5+ is an excellent move. Here is the position just before white played it:

click for larger view

How can we spot 31. Nf5+ from here?

I think a lot has to do with the questions we ask ourselves when we analyse a position.

If we ask the question: "What move should I play here?", our minds will often give us a list of safe-looking moves. We will usually subconsciously filter out any moves which instinctively look wrong or unsafe.

But suppose we did manage to list 31. Nf5+ in our first scan. What question do we ask then? Most folk ask the question "does this work?" And that is okay as questions go, but it is all too easy to take fright at the fact that f5 is twice defended.

Instead of "what should I play?" and "does move X work?", we probably ought to substitute three more powerful questions:

"What do I want to achieve?"

"Which moves achieve what I want?"

And for each of the moves in our candidate list - "How can I make move X work?"

From the diagram, white would really like to shift the black king out of the way so that he can play Qxf6. That answers our first question, at least as an initial hypothesis.

Then our second question leads us to the seemingly impossible 31. Nf5+ as the only move which shifts the Kg7.

Finally our third question makes us persist with 31. Nf5+ instead of giving it a cursory once-over and reject it. There are three variations to consider - king moves is followed by Qxf6; exf5 allows Qxf6 (as in the game) and gxf5 gives us Qg5+ and Qxf6.

It's all about asking the right questions!

<kramputz> You are entitled to your opinion of course. Not everyone will like the stuff I write. So feel free to use the ignore button. I won't be in the least bit offended.

37...Qd4 38.Qxd4;
37...gxh5 38.Rg1+ mate.

First for the dramatization, and second for the insight on 31. Nf5+. I found the puzzle easy to solve, but when I played through the game, my reaction to 31. Nf5+ was similar to <Suffering Bruin>'s: how do I "see" a move like that during a game?

a Tuesday. Love it. Sometime even more than a Monday.

37. Nh5! and Black has no defence.

37...gxh5 38. Rg1#

and after all other moves: 38. Qg7#

Thanks again for an interesting storyline and also your insight about asking the right questions.

To see a move like 31.Nf5+, I use the "check, capture, and threat" method for considering candidate moves. But for a move like that to remain on my candidate list I do a "safety test" to determine if the move is safe. If the piece can be captured without a following "check, capture, or threat" then the move is likely unsafe and time shouldn't be wasted analyzing it. As you pointed out, the critical variations stemming from 31...gxf5 and 31...exf5 suggest it is safe and therefore worth considering.