Wednesday, April 30, 2008

The Trap

There are 3 kinds of scans I try to improve:
  • Trap
  • Converge
  • Targets
In this post I want to talk only about the trap.
The following diagram is quite well suited to explain how the exercise works.
In the problemset of Chess Tempo there are a lot of mates. I consider a mate to be a special instance of a trap. You trap the king. A mate like below took me quite some time since there are a few initial moves which look promising but which are not.

White to move.

You don't know beforehand that this is a mate, but it isn't likely that you can win a piece.
Solution [1.Qxg7+ Kxg7 2.Rf7+ Kg8 3.Rxe7+ Kf8 4.Ng6#]

It isn't important to spend much time to find the solution but it doesn't hurt either. There is no objection to look it up. The exercise starts when the solution is known.

The input is the position at hand.
The output is the endposition with the mate.
The narrative describes in words how you arrive at the mate starting from this position.
How the white pieces work together to box in the black king. In this kind of mates typically the six squares f8, g8, h8, f7, g7, h7 must be covered. Imagine from the startposition which white piece will be responsible for which of those 6 squares.
The goal is to see the characteristics from the endposition in the startposition.

You must ask yourself the question, what characteristics do I need to notice to be sure that I recognize the solution immediately the next time I get to see this position?
It is not so much about visualising the move sequence, but to relate the input to the output simultaneously.

For this post no mice were hurt.


  1. My verbalization. It's a problem position, so I should look for something forcing. In a game, I hope I can think of a queen sac. The most forcing move is Qxg7+. No alternative but Kxg7. Now what? The only indefensible check is Rf7+. Hmm, now I notice that the Nh4 prevents Kg6. So Kg8 or Kh8 is forced. At g8, there is a bishop discovery; at h8, the knight comes closer with Ng6+. If Ng6+, it covers both h8 and f8, while the bishop hits g8. Now we just have to get the rook out of the way. On e7, the rook is out of the way and defended by Ng6. So after Rf7+ Kg8 Rxe7+ Kf8 Ng6# as in your solution. After Rf7+ Kh8 Ng6+ Kg8 Rxe7# still works. I don't see the ending positions unless I first find Qxg7+ and then concentrate for about 3-5 minutes. Certainly, it's not instantaneous.

  2. Soap,
    do you have any idea what is needed for you to see it faster in the future?

    Was the queen sac purely based on the the fact that you knew that this is a problem context?

    I focussed on the radiation of the pieces and how they cooperate in the hope to see it faster in the future.

  3. i have a draft post about something along these lines, it's about backwards planning. i know where i want to go, i just have to devise a way to get there. as for speed, i don't care if it takes me longer than everyone, if i see it, that is what matters most to me. so, i know i want my rook here and my bishop there and his pawns here and there...if i can make that happen, i then know what i need to do. this comment probably doesn't make sense, sorry...

  4. CL,
    it makes perfect sense. For backwards planning you need to see the future position. To see the future position you must see its characteristics in the current position.

  5. Perhaps this is saying the same thing as the radiation of the pieces. When I think now in retrospect at the combination, following Qxg7+, I see a wave front composed of the squares White controls starting with f6 and g6 plus Black's pawn on h6. Next, f7 controlled by the bishop and g7 and h7 controlled by the rook arriving on f7. Then back rank squares g8 controlled by the bishop, but blocked by the rook, h8 and f8 controlled by a knight arriving on g6. There is some movement required to see the rook moving f7-e7 so that the bishop is unmasked, then I add e8 controlled by the rook and e7 controlled by the g6 knight. I can see it as a succession of interlocking squares that catch the king in an airtight net. Seeing the squares lock together seems to take less time than watching the pieces move about and then radiating each one, but again I say this in retrospect. Insecurity about the accuracy of my calculation (and punishment at Chess Tempo for inaccurate lines) tells me that my slow original method is more careful. But there are some days when my "apprentice" works surprisingly fast and I can't explain it.

    I utilize Chess Tempo in the Standard mode with little concern for time (not uncommonly going 5-25 minutes for just one problem if I'm not satisfied I've nailed it). I think it has honed accuracy and sticktuitiveness, if not speed. Also, some ideas are not so easy to see, so it pushes me to look more thoroughly for the strongest idea, which I think helps my scanning somwhow.

    I think it goes without saying that Qxg7+ is much easier to find in a problem setting. During a game, as long as I have thought "Wow I've got a tremendous force advantage around the enemy king," sacs are likely within my ability to conceive of them. If my pieces aren't clustered about the king, yet such a combination still exists, I think my probability of conceiving the sacrifice is low. Consider my post on Hikaru's Immortal All the pieces are so far away, Qxf2+ is nearly impossible to see.

  6. there's no way I could see the queen sac in a real game, but what's stopping the Ng6+ Kh7 Qe4 mate? I'm drunk (1st may, huge holiday in finland) so I'm probably missing something really basic. but what??

  7. WW,
    Sounds like you having a good time.
    Qe4 doesn't give a check. Your line is mate in 6, while it can be done in 4. But this post wasn't about the solution but the way to improve the next time.
    Have another beer.

  8. Soap,
    if my hunch is right, to improve it is necessary to see in "protospect" (advance) what you now see in retrospect. It gives you the overview which you lack when you look at it move by move.

    That is the idea behind the exercise I proclaim. It doesn't seem all too far from what you are already doing.

  9. okay, I just checked it with fritz, and there's no serious moves to defend against the Qe4 mate. first Nf5 giving up the knight, then Qg8 giving up the queen.

    it's interesting how my mind refused to see those moves even after you said it's mate in six. I just kept looking for reasonable moves, ones that actually do something, and completely missed the silly Nf5 and Qg8.

    the Qxg7+ mate still seems hard for me to visualize, even though I know it's there. my knight/bishop vision must have degraded since the time I drilled KNB mate extensively, and I guess I should go back to it for a while. maybe KNB is something that should be drilled every now and then, regardless of how well you know it. like singing scales is basic daily training for singers, no matter how good you get.

  10. WW,
    your fast improvement in chess must be related to your "tunnelvision" which prevent you from seeing silly moves.