Friday, August 17, 2007

From word to sentence

When there are thickets of variations in a position, there are parallel lines and serial lines. I always thought that parallel thinking would be the most troublesome in terms of overload of the short term memory. Since the short term memory is made for sequential work. In my last post I surprised myself by saying that it are the long serial lines that intertwine the basic tactical elements that cause the problem and not the short parallel lines. After two days of investigation of this statement I more and more become convinced that it is true.

When you call a basic tactical element a word, then the line is called a sentence. You have to learn to read the sentence. The grammar dictates how the words are intertwined.

What are the topics while attempting to read a sentence?

When there are 9 basic tactical elements in a row, you must be able to see how the position will look like at the end of the line. That will take considerable exercise.

The moves from one tactical element tend to interfere in the mind with the moves from a previous tactical element. In fact that is a visualisation problem. This serial interference is far worse than parallel interference.
Another form of interference is that you know what has to be accomplished, but that you have trouble with the move order.

Pattern recognition.
You must recognize all tactical elements in the sequence.
The construction of narratives helps to formulate to which tactical element (pin, fork, skewer etc.) the moves belong. This makes that it becomes conscious, which is crucial for learning and this eases the unconscious pattern recognition in the future.

The patterns have to be so familiar that it isn't necessary to count to know how much you are ahead in material.

Forcing moves.
A tactical element must be forcing to a certain degree. If it is not forcing, the opponent can do anything as an answer. You must be very aware of the limited moves your opponent can answer, since that are the moves that needs investigation. That way you get rid of a random list of candidate moves, which is characteristic for the amateur, in favour to a shortlist with moves that must be investigated.

If you end the sequence with two pawns up plus the bisshoppair for the exchange you must know if it is worth it.

I can't get rid of the feeling that there must be something more. I hope to find out.


Problem 388

Black to move and win.

I will not treat all lines, it's just to give you an idea what i'm talking about. The basic tactical elements are red.
The first thing that catched my eye was the pawnfork at d4. To avoid the loss of wood, white has to move Nxd4 as an answer. All of a sudden a discovered attack with Bb7 and Ne4 against g2 shows itself. Since the knight on f4 is hanging anyway, a knight sac as decoy for the king to g2 as preparational move arises.
The move order has to be corrected:

1. . . . Nxg2
2.Kxg2 d4

This should be the first steppingstone for visualisation (Tisdall).
If white does the "normal" move 3.Nxd4 then 3. . . . Nxc3 restores the material balance by good position for black.
In stead of default natural reaction you have always to be alert by more forcing counterattacks. In this case 3.Nd5 or 3.Nb5. I just treat

3.Nd5 Rxd5!

This second temporary knight sacrifice prevents the loss of the exchange by the skewer 4.Bxb6 further saving d4 and b6

4.Qxe4 Rxe4 discovered attack against invasionsquare e2, pinning knight f3
5.Qh4 Re2+ invasion with double attack
6.Rxe2 Rxe2+ forced trade
7.Rf2 Rxb2 exchange sacrifice
8.Rxb2 Bxf3 trade
9.Kxf3 Qc3+ double attack regaining material.

Black ends with two pawns up and white's king is vulnerable.
I have not treated all lines but in these lines alone you can count 14 basic tactical elements. It took me 3 days to work everything out. I hope it's just a matter of exercise to learn to do this within an hour or less.

The question is, how far did the grandmaster look ahead? In the openingsposition blacks knight is hanging and it cannot escape, really. So the first move Nxg2 is not so difficult to find. After move 3 black has his knight back, so that is a natural steppingstone. But in order to decide between 3. ... Bxd5 and 3. ... Rxd5! all the lines must be worked out till quiescence. I can't come up with a narrative why Rxd5 is better other than the concrete variations.

This is a peek in the grandmasterly world of calculation and I will continue to study problems in a grandmasterly way. Did I already mention that it is very demanding?


  1. Ah yes.

    My problem now is that on some moves I am starting to see things in more complexity (not the level you are seeing, but starting to see multiple motifs, visualizing them a couple of moves ahead, helping to guide my strategy). The problem is, it takes a lot of time off my clock. So I'll do a really impressive analysis for one crucial move, but sometimes that's like 20 minutes off my clock. Then, in time trouble, or because that initial think tired my brain, I end up making a really stupid move (like not seeing the correct defense in my recent game).

  2. Blue,
    if I had to make an assesment, I would guess that thorough knowledge of the endgame would add an 100 ratingpoints to my rating. And a study of the middlegame mechanics might add another 50 points. Openingstudy maybe 20 points. With 170 points extra I would have a respectable rating of 1900 or so, exactly the maximum what you would expect from somebody of my age, who learned the game later in life. These extra 170 points would be based on knowledge.

    But besides this, I belief that the rest of the difference of a grandmaster and me lies in exactly this, the ability to calculate long serial lines well. And that is why I focus on this aspect.

    Knowledge is very democratic. Everybody who is willing to study can get it. Often in fairly little time. But it is the skill to apply knowledge what makes the difference. And that skill is not easy to obtain at all.

    You are right, the skill must become swift and easy. Otherwise it will take the minutes away from your clock, and it takes away your mental energy, letting you make mistakes.

    As MDLM said: the skill to see one ply more ahead is more worth than all the positional knowledge in the world.

  3. But besides this, I belief that the rest of the difference of a grandmaster and me lies in exactly this, the ability to calculate long serial lines well. And that is why I focus on this aspect.

    Do you believe that a grandmaster may be more familar with a great many more positions and and knowledge if they are plus,minus or equal? Knowing what is known and requiring less need to calculate? I wonder if there is some pragmatic knowledge of how to compete in tournament play time control etc that may enter into it as well.

  4. Tak,
    no, I don't think that the fact that they know more positions is crucial, but the fact that they know the 10 (or so) basic tactical themes better, at a higher level, that is. They don't know more words, necessarily, but they know the grammar of which we have very little idea. They can make far more correct sentences with the same simple words. It is the quality of their pattern recognition that make them recognize a few themes everywhere. That's how the reach quantity.

    Again: they know the higher level cognitive rabbit very well, hence they recognize rabbits everywhere, even where we wouldn't think of a rabbit.

    Since they know the basic themes so well, they have to calculate less.

    About a time-use strategy:
    When there are forcing moves, you must calculate untill quiescence, otherwise it is just a gamble. So if you are not able to calculate everything untill quiescence within the available time, a time-use strategy equals to a gamble-strategy.

  5. What a position... :-O

    Your comments make me think of music. It takes skill to play an entire piece by reading music or memorizing it, and yet improvising a complete composition requires another level of mastery.

  6. King,
    nice to hear from you again! All is well I hope?