Saturday, November 15, 2008

Simple Forcing Variation

The following tactic is fairly easy to calculate, yet it gave me some trouble. I'm finding that, when calculating forced variations, I have more difficulty when I have a choice of moves than when my opponent has a choice of moves.

White to play:

4r1k1/ppn1bppp/q7/7Q/1nN1P3/1B1P4/PP5P/KN4R1 w - - 0 1

Black to play has a straightforward mate in 2 starting with 1...Qxa2+, so White's play either has to interfere with this mechanism, or be with check. Moves like 1.Nca3 or 1.Na5 don't lead to anything. That leaves two checks: 1.Rxg7+ or 1.Qxf7+. The former check doesn't seem to lead anywhere either (e.g. 1...Kxg7 2.Qg4+ Qg6).

1.Qxf7+ looks like an "obvious" queen sac, because after 1...Kxf6 the knight can move to e5 or d6 with a discovered, double, check. But which knight move?

Correct is 2.Ne5+, because on either 2...Kf6 or 2...Kf8, 3.Nd7#.

However, when I tried to calculate this as I would in an over-the-board game, I would stall at this point. I was seeing the king slipping out of the mating net with 2.Nd7, and didn't immediately see the mate after 2.Ne5 Kf6. That was enough for me to second-guess the entire line and try to find ways to get other first moves to work. In a real game I probably would have bailed out by playing a knight to a3.

This demonstrates both quiescence errors (not calculating out until all checks, captures and threats are spent) and not "thinking like a tree" a la Kotov (calculate each branch of each line only once). And yes, Kotov's technique has been criticized by others, but in general (especially for simple problems like this one) it's an ideal worth striving for.


Anonymous said...

I have to agree with you...I thought that Ne5 was better than Nd6, but I kept on thinking that the Rook was required after Kf8, rather than seeing that the Bishop's diagonal extended over f7.

Very nice. I think I would have played ne5 anyway and trusted to a perpetual check ( which is what I kept seeing )


chesstiger said...

I think i would hesitate to play Qxf7+ because i probably will miss that after Ne5+ it's always mate.

It's also not natural this Qxf7+ because you give a full queen for a little pawn. Sometimes this bad habit of pointcount hinders me, especially when the difference is so much.

Anonymous said...

I saw Nd7 mate when the K moves to f8 but missed it when the K moves to f6. Arrggh!

I'm not sure I see how Kotov's calculate each line only once applies, how does it?

Grandpatzer said...

Re: how Kotov applies:

As I read Kotov, each branch would be analyzed out to a conclusion, then the next branch, etc., exactly once.

My stubby branches weren't analyzed out to quiescence (no checks, captures or threats) and I was bouncing between candidate moves, rechecking failed lines.

Anonymous said...

Re Kotov.

Ok good point, this is a good situation for the Kotov thinking. Maybe listing all the Black King candidate moves would help to find Nd7.

I just read Tisdall's take on Kotov in Improve Your Chess Now. Very interesting.

Dragan Glas said...


Mates and material-winning combinations involving knights (and intermediate pawn moves) are difficult to spot - as Kramnik found out to his cost!

This is mainly because most mate patterns involve the linear pieces (bishops, rooks and queens).

And there lies the key - patterns.

If one can practice patterns involving knights, this can help.

Backward moves with knights are particularly difficult to spot within combinations.

If you don't yet have the pattern-recognition entrained into your brain, one method is to consciously note which squares the linear pieces are covering (bishop and rook in this case) and then note any squares covered by pawns (e4 here).

That leaves the three squares - e5, f6 and f8 - which should immediately suggest the Nd7 move.

Granted, if the knight weren't already on one of the key squares, say b8, it'd be more difficult to spot!

Kindest regards,

Dragan Glas