Sunday, May 11, 2008

Cumulative analysis

Confusion. . .

In order not to confuse you I had better not written this post. It is about a nagging idea which I can't get rid of, so in order to set me free I share it. Sorry, I don't mean to disturb you.

I'm completely happy with the scanmethods that I described in previous posts. And that is what I am training right now. In fact the method is a form of extended microdrills, with more practical value added.

The idea that I can't help thinking about might not be possible or practical at all. Yet there is a certain theoretical beauty in it that causes me to think about it.

The scan methods as described are useful in any position. I mean that you can break in in the middle of a game and start the diagnosis with these scans. It is quite useful to be able to do that. But that is not how we play. We play move by move, which means that theoreticly we can diagnose the game in a cumulative way. In order to do that, all consequences of a single move must be seen. That is quite a task. Maybe even too big to be practical. But it is the only way to prevent your positions from being accidental. I will try to explain it with a diagram.

White to move

1.Ra7+ Ke8
Common practice is that I treat the new position as a new position. Which it is not. Let me focus at move 1. ... Ke8. If my diagnosis was up to date so far, then I have to look only at the cumulative effect of Ke8. That is to say: the black king leaves f7, which has certain effects, and he is put on e8, which has certain effects. An effect of the latter for instance being the interruption in the communication between the rooks. With cumulative analysis you have only to worry about the current move, since the cumulative effect of the previous moves is already known.

That is not how it works with me now. Since I don't look at the exact effect of every move, I must treat the new position as new. That is to say, I have to make a full scan of the position, as if I have never seen it before. If I'm lucky those scans reveal that the rook on b8 is no longer protected. In the position above it took me about a minute to see that. Simply because I was looking at other parts of the board first.

I wonder if it is possible to enlighten the task of cumulative analysis by making it a habit. That would change the feeling that I now often have: that a position befalls me accidently. That the features of the position seem to appear out of the blue.


  1. This isn't unclear, but is cool. Heisman talks about it, suggesting you only need to evaluate what has changed since your last evaluation.

    I can't remember if it was him or someone else I read in this context who said the following two things will change:
    1. One square will now be empty that wasn't.
    2. Another square will now be occupied by different material (if it was empty it will now have material, with a capture it now has different material of your own).

    When I first saw someone say it (a master of some sort), I was like 'No shit, that doesn't help me! This is trivial.'

    But it is strangely beautiful and abstract, I appreciate it more the more I've played. You can focus on changes that result from the changes at the two squares (e.g., discovered attack when a square is left empty, a pin due to the bishop being on a new square).

    Getting back to the point, if you have been looking at a game from move 1, playing it, then to evaluate the new move you can thoroughly look at the implications of the change in the position based on 1 and 2. That was the point of the person I was reading--if you have been evaluating well the whole time, have a good sense of the board, then you should focus in especially hard on the changes wrought by 1 and 2 for the last move.

    As I improve (ever so slowly) at chess, it is these trivial axiomatic details that I realize I have trouble with in real games. And understanding that changes at these two little squares can have drastic repurcussions for the flow of the game, that it is actually incredibly hard to really deeply understand the implications of such a simple change of material from one square to another.

    Of course, just focusing on two squares is a huge mistake, and that is not the point of the person's observation. Part of the point is that these simple little changes at two squares have crazy complicated interactions with the entire board, often changing the game from =/= to +-, and the better you are at noticing the changes (rather than evaluating the position anew) the more efficient you will be in real games.

    It also sort of assumes my previous analysis didn't suck. Perhaps my queen has been en prise for five moves, neither person noticing. In that case, it probably isn't the implications of those two squares to be focusing on, but looking at the board like a baby (this is the point of blunderchecking, to make sure you haven't gotten lost along the way).

  2. I already investigated some of this stuff in a previous post. I have tried it but it is quite a workload per move. If you have 6 candidate moves of your own and 6 of your opponent for each of your moves, you need to play a corresponce game. And be very disciplined.

    Only if the workload can be diminished drastically by making it a habit, this might work. What remains is the valuation. You move a pawn, it clears two diagonals, but what is the value of those diagonals? Can you make use of it or your opponent? etc.

  3. Did I miss anything in the following ?

    Pins, and forks and hanging pieces.
    Skewers, Invaders and back rank weakness
    Kings without air and Defence not there
    X-ray Attacks and guards that get wacked,
    Pieces in a row and Queens with no place to go.
    Attacks discovered and Kings uncovered
    Discovered Checks and Pawn structure wrecked
    Open line through and defenders with too much to do.

  4. Tak,
    LOL. The problem is not a lack of possible scans, but a lack of time to learn them all. You have to find out which scans are effective enough to devote your time to.

  5. Bdk. I can't find the right music to put it to. The only thing I can think of is Misty Mountain Hop and a Few of my Favorite things. 8)