Friday, July 20, 2012

Doing something


.
.
.
.
In the position above you can place the rook in three different kind of places where it performs three different kind of actions:
  • On Rg1: Nothing.
  • On Rf1: Taking away squares from the knight.
  • On Rd1: Attacking the knight.
Full potential.
On every square where you place the rook it attacks 14 squares. That is its full potential.

Mobility.
If you place the rook at f1 the possibilities of the knight are diminished by two squares.
At the same time the full potential of the rook is limited by two squares. That leads to its actual possibilities. Its mobility. Full potential - limitations = actual possibilities = mobility.

Attack.
If  the rook attacks the knight, there comes a tactical element into the game. The problem though is that you can chase the knight untill the cows come home, it simply moves away.

Tactics naturally flow from good positions.
Attacking is useless as long as the enemy has enough posibilities to meet the problem. In order to make an attack succesfull an element of force must accompany the moves. The moves must be forcing. As long as the opponent has enough possibilities, there can be no forcing.
This gives a clue what positional play should be aiming for: increasing the own possibilities while dimishing the possibilities of the opponent.

How to limit mobility.
Limiting possibilities is often mutual. If the rook deprives the knight from 4 squares, the knight mutually deprives the rook from the same 4 squares. How can one make progress in this area then?
Here is where the value of the pieces kicks in.
See diagram below.

















.
.
.
.
The only reason that pawn e6 limits the mobility of the bishop is that it has a low value. If there was a knight on e6 or a rook, the possibilities of the bishop weren't severely diminished since black can simply trade the pieces off. So the main method of positional play is: diminish the mobility of the enemy pieces by containing them with pieces of a lower value.
This makes the modest pawn the piece nec plus ultra for positional play.

Another method of positional play is outnumbering. For this you can use pieces of equal value.
See diagram below.

















.
.
.
.
White outnumbers black on b7. He can put a piece on b7 and keep it there. If black trades the pieces off, white will end up with a rook on b7. This is the method for invasion squares and outposts. (allthough you don't see it b8 st protected by an invisible piece that I have left out to avoid complications. But you get the idea :)

A little side step.
What is the static value of a piece made of? If the rook has a value of 5.00 pawns and the bishop a value of 3.00 pawns, where does it stemm from? Somehow these figures express the actual activity of these pieces. The statistical average of what a piece actually does in a game. Since I now have defined what "doing something" means, it is worth to attempt to express the relative strength of a piece by means of the squares it has an effect on. To make a calculation based on potential activity, limited squares, mobility and so on. This offers the possibility not only to express the statical value of a piece, but its dynamical value too. The natural unit of such calulations will be "amount of squares" in stead of the arbitrary centipawns. Only this way we can hope to come to a common base of all the bonusses and penalties found in mainstream chess programming. Only then we will be able to define a theoretical support for these parameters, in stead of the empirical foundation we use to day.
But that will be by no means easy. And it is not what I'm after at all. Yet it must be said.

Summary.
Positional play is aimed at improving the relative mobility of your pieces. The methods are containing pieces by pieces of a lower value and outnumbering on invasion squares in case of pieces with equal value. When the mobility of the enemy pieces has become below a certain threshold, tactical combinations will start to manifest themselves.

5 comments:

  1. A few minutes after completing my post regarding flexibility, I read yours concerning mobility. I rather suspect that you, too, have been reading Dan Heisman, The Elements of Positional Evaluation.

    Am I correct?

    ReplyDelete
  2. No, I haven't. But I tend to pick what's in the air:)

    ReplyDelete
  3. Ehh,.. that was what i meant in my first reply to your last post . That anaylsis can be done by a slightly modifyed "move(list)generator".
    Some squares could be weighted higher: Squares in the center, in front of your pawns, close to the kings, in the teritory of your opponent...

    many positional rules are just "this"

    ReplyDelete
  4. @Aox,
    I know, I know. But that's the pragmatic approach.

    Chess engine programmers don't realize it, but now human is no longer a match for the computer, its time that they search for a new goal in life.
    Due to inertion they continue to do what they always did, i.e. programming of engines in a pragmatic way. The goal is now to let play engines against other engines. How silly is that?

    My plea is that they shift to another area: give the pragmatic empiric approach up and try to give matters a theoretical sound fundament. Which means: start from scratch and make understanding and gain of knowledge the goal.

    I'm well aware that I ask too much. Programmers are pragmatists, not theorists. But now their approach to chess engines has become obsolete, maybe I can inspire a few to think again.

    But this all is not important at all. By thinking about these things I make progress myself. That's the only thing that counts for me.

    ReplyDelete
  5. To write a better chessengine is a sport like to make better moves at chess. But creating a better Chessengine is a indication of beeing a better programmer, getting a higher Elo is only indicating a better chessplayer.
    There is "nothing" as obsolete as chess :(

    ReplyDelete