Monday, March 21, 2016

Annihilation of the defender

I  use a simple counting system to see if I dominate an attacking square: #attackers - # defenders.
That simple system "as is", is in fact useless. The moment I look at an attacking square, I must look immediately to its defenders and see if I can annihilate them. Only the numbers after annihilation of the defenders are useful. Take a look at the following diagram:

Black to move and win
r5k1/5p1p/3p2pQ/1ppP2rP/4P3/5Pq1/1P3R2/1BR4K b - - 1 1
Solution

 Attacking squares:
  • f4 +1
  • h4 +1
  • h5 +1
  • g2 0
  • g1 0 
  • h3 +1
  • f3 0
 As you can see, the numbers give no clue where to start. What you really want to know, is if you can dominate a square after you god rid of the defenders. There are three main ways to get rid of a defender:
  • Capture
  • Deflect (defender is overworked)
  • Attack
There are other ways to get rid of a defender, like blocking it, but since you need to preserve the initiative, the frequency of these solutions is low. And I want to use a thought process only for frequent occurring situations.
What are the defenders:
  • f4 no defender
  • h4 blockable by rook
  • h5 defended by queen
  • g2 defended by king and rook
  • g1 defended by king and rook
  • h3 blockable by rook
  • f3  defended by rook
1.... Qf4 is already a winning move, but it is not the best move.

h3 and f3 give you a clue: the rook on f2 is overworked. From that the moves can be easily determined. 1.... Qh3+ 2.Rh2 Qxf3 3.Rg2 Qxg2#

The thought process can be summarized as:

1. What's the position about?
  • Gaining wood
  • Checkmate
  • Promotion

2. What are the targets?
How to remove the defenders of the targets:
  • Capture
  • Deflect (defender is overworked)
  • Attack
3. What are the attacking squares?
How to remove the defenders of the attacking squares:
  • Capture
  • Deflect (defender is overworked)
  • Attack
Step 1 and 2 flow rather naturally. Step 3 usually reveals the clue to the position. Once the clue is determined, finding the moves is not so difficult, most of the times. The problem is to learn the mind to process step 3 automatically.

Investigation of more than 100 positions at CT convinced me that mastering step 3 is paramount. The exercises I asked Lain to develop fit exactly into step 3. Which is no coincidence, of course.



14 comments:

  1. Do you refer to the "simpler"/"easy" FAC-based versions of IAPP/IABP?

    I extracted these positions some days ago: http://puu.sh/nOK0D/c0e733d4bf.txt

    Then I stopped and I didn't test anything. I thought that mixing FAC with another subtasks its not a good idea -Aox said that-.

    ...Should I continue extracting positions?

    ReplyDelete
  2. Simpler IAPP and IABP as we agreed upon will work for persons who have mastered FAC first. I think these simpler IAPP and IABP will proof to be improvable, and that we will need the exercises sooner or later. If you don't have to invest too much work, you probably should continue. It is up to you to decide. We are in uncharted waters, so we can't be sure beforehand that what should theoretically work, will work indeed in practice. I agree with Aox that we probably need more simpler exercises too and that those will probably will work better. If working out the simpler IAPP and IABP is too much work, you should wait for a next post of me, where I will try to work out all exercises we need in their most simple form. The very chess atoms. But then again, we are in uncharted waters.

    In that post I will look for the chess atoms that will be beneficial for all tactics, not M1-x alone (I always looked at that bigger picture). There must be a close connection to the exercises and the thought process I'm describing here, since the chess atoms should help to speed up the thought process.

    ReplyDelete
    Replies
    1. http://spookiecookie.weebly.com/files/theme/chessex/IAPP.html
      http://spookiecookie.weebly.com/files/theme/chessex/IABP.html

      Testing time.

      Delete
  3. as i see you get closer to the system of chuzakihn, he is counting all x-ray attackers but not x-ray defenders and substract the erasable defenders. Squares and pieces with a count <=0 are critical.

    I think the thinking process at a puzzle should start with counting material. This way you touch each piece at least once with your thoughts and this info is important.
    You need to count your material at the end of a combination blindfold too, so its good if you can do that with ease.

    After your point 1 you need a "Case of"
    If the position is about gaining wood you need to decise: which pieces might be won and then how to win them
    If the position is about promotion...

    ReplyDelete
    Replies
    1. A thought process is highly personal. Habits that are already automated should be left out, since they already take care of a certain aspect. Analysis of the failures at CT provide the clues which aspects need improvement. Those aspects must be accounted for by adding them to the thought process.

      Delete
  4. @Lain i do a lot of FAES-e but the optional 00 and 000 are not helpful in my eyes. I would suggest to take positions without casteling or simply to ignore that move.
    It is "interupting" to have to look at the writings if casteleing is possible or not.

    ReplyDelete
    Replies
    1. Updated, castling moves are removed from the solutions in FAES-e.

      Delete
  5. Interesting perspective on the "removal of the guard" theme, it's certainly helpful to consider deflections and overloaded pieces as part of that process, rather than just outright captures.

    ReplyDelete
  6. Looking a the puzzle:
    White is up a piece vs a pawn. ( which means that if the puzzle is about material black would have to win something like a rook but the eye catching Rh5+ Qxh5,Pxh5 dont work because of Rg1 )
    The number of attackers of the white king is 2, the number of defender is 1( + 1/2), this puzzle is about mate
    So all the weak pieces are not interesting
    Now we need a mate pattern, no check at the backrank is possible...

    By the way this puzzle is a nice excample of cct
    There are not many possible checks... here a strict CCT is superior

    In my eyes the analysis of attacking squares is at this puzzle wasted time

    ReplyDelete
    Replies
    1. The position is merely a vehicle to talk about the necessity to look for guards and their removal when looking at an attacking square. That's a habit I want to cultivate. As usual the actual position distracts from the things I want to talk about. It is not easy to find positions that show subjects in an isolated fashion.

      Delete
  7. Master Charles Hertan recommends (in his book Forcing Chess Moves: The Key to Better Calculation) investigating FORCING MOVES FIRST in tactical situations. I start with the King: does the King have any moves available, or is it "in the box"? "King in the box" draws my attention like a moth to the flame. If it is already confined to the "box" AND there are at least two pieces available for a direct attack on the King, then there are favorable preconditions for a mating attack. One piece can be used to directly attack the King while the other piece protects the attacking piece, and/or controls parts of the "box" surrounding the King so that it cannot escape. By using forcing moves, you can (in essence) "force" the opponent to play his piece(s) to squares that you choose. The White King IS in the "box" AND the most forcing move in this problem is a check with the Black Queen, because the Black Rook prevents the White King from escaping from the "box" by moving to the g-file. 1. ...Qh4+ (forcing) 2. Rh2 (forced) leaves Black with no immediate followup forcing move. 1. ... Qh3+ (forcing) 2. Rh2 (forced) Qxf3+ (forcing) 3. Rg2 (forced) Qxg2#

    It is the combination of "King in the box" and available forcing alternatives that grabs my attention in similar situations. It is just one of many mental "shortcuts" that I have accumulated over the years. It is NOT part of some formal thinking process. (I am aware of the aphorisms "Patzer sees a check, patzer gives a check" and "Always CHECK: it might be mate!" Not the same thing at all.)

    As AoxomoxoA pointed out, there is a "diversion" in looking at capturing the White Queen in exchange for the Black Rook. I rejected that option as a FIRST attempt because of the geometrical line from g1 to g8, on which the Black King and Black Queen both reside, with either of the White Rooks available for pinning the Black Queen to the Black King. This is where Aox's previous post about being aware of the material balance (evaluate the material balance FIRST) comes into play: an even exchange of pieces favors White.

    I know it sounds counter-intuitive to consider (and reject) this "obvious" line of play based on geometrical considerations along a line (g1 -> g8) when there are two "obstructions" blocking that line at present (the Black Pawn at g6 and the Black Rook at g5). However, it doesn't require much board "sense" to realize that capturing the h5 Pawn removes the Black Rook, and capturing the White Queen removes the Black Pawn, opening up the g1-g8 line. It doesn't eliminate the possibility of trying that line of play but it does give an "intuition" that something about that line of play might not be the best FIRST idea to investigate.

    Back to lurking mode. . .

    ReplyDelete
    Replies
    1. The problem is, that I'm easy seduced by going into Trial&Error mode. All this CCT and forcing moves, causes me into T&E-mode. T&E-mode is forward thinking.
      See http://temposchlucker.blogspot.nl/2006/07/trial-and-error.html

      What I'm looking for, is ways to think backwards. Once my attention is drawn towards f3 and it's guard, it is easy to reconstruct the moves. The method I describe, takes me to f3, albeit not in the fastest possible time.

      What I do, is solving a series of problems at CT using my thought process. I make very little errors this way, but I consume a lot of time. Later on, I will have a closer look at the time consuming problems, and look for alternative methods like you describe to speed them up. First doing it correct, then speed it up.

      Delete
    2. I understand your point about CCT and T&E being "forward thinking." If there is nothing more than aimlessly trying out various CCT or T&E "forcing" moves at random, then I agree totally with your assessment.

      "When you don't know where you are going, then any road will take you there."

      On the other hand, when a very clear "signal" (such as "King in the box") is present, I submit that the goal/target directs the attention to and limits the attention to the "forcing" moves which achieve that goal, thereby narrowing the T&E aspect of choosing/trying out moves in the mind.

      Dr. Lasker was quite specific about the distinction between the two archtypically different mindsets of positional and combinational players. (Please note that he synthesizes those two types in the "master" player.) The combinational player uses a CCT and/or T&E approach, trying the violent, forcing moves FIRST and then checking to see if a position is reached which is advantageous. The positional player sees a desired end position FIRST, and then checks to see if there are moves (or series of moves) that will bring that envisioned end position into being. The "master" uses both approaches in a synthesized process.

      I look at it this way. If I can discern something in the position that gives me a "signal" as to a desired goal or end state, then I try to see if there are forcing moves that bring me closer to that goal or end state. The "forcing" moves may appear to be CCT or T&E, but are constrained by the goal to a much smaller subset than purely T&E. Working out the intermediate positions and the most effective ordering of moves to arrive at a goal state does not constitute T&E, IMHO.

      "First doing it correct, then speed it up." I agree unreservedly with that statement!

      Back to regularly scheduled programming. . .

      Delete
  8. I think a thinking proces is a pseudo code which should work with any puzzle ( of a given type ).
    So this puzzle is not a bad example.. our "code" has to work here too.
    I suggest to find a reasonable pseudocode for this puzzle and expand it step by step, puzzle by puzzle.

    ReplyDelete