Wednesday, August 01, 2012

Tactical technique and geometrical patterns

Yesterday I dug up an e-mail out of my spamfolder from a "real chessplayer" from the Ukraine. His name is Alex. He has some very interesting things to say about chess improvement. I intend to make a post of the e-mail, but in order to do so I will have to translate it from Ukrainian english to dutch english. His english and my chess have a lot in common ;)
Please give me some time.

When you have to make a move in a complex tactical problem, the endposition is usually not directly clear. Often you go forward one move at the time untill you recognize a geometrical pattern. The natural place of a geometrical pattern is at the end of the branches. When this killing pattern isn't in sight yet, you must base your next move on other considerations.

Of course there are geometrical patterns around all the time. But only when a branch reaches its end, the patterns all of a sudden become dominant. It contains a roadmap to the future, laid out in front of you as a geometrical pattern. When you encounter a pattern that is not at the end of a branch, this pattern is just another consideration for your move.

80% of the time during the solution of a tactic, you will have to base your move on logical considerations. Geometrical patterns might or might not play a role in it. It turns out that I am terribly bad at these logical considerations. And I feel so free to extrapolate that to the most of the readers of this blog: most of us suck at logical considerations.

When we haven't reached the geometrical pattern at the end of a branch, we will have to think per move. What must be the goal of this move, what does it accomplish? A lot of precision is needed here. Sometimes subtlety. This I'm inclined to cal "tactical technique".

Usually we have no shortage of geometrical patterns in our database. Especially those of you who have some serious tactical training under the belt. The main cause why we suck at tactics is that we lack "tactical technique". The training I proposed in my previous post is designed to solve this lack of technique.

Let me give an example.
You can find the complete problem here.

White to move. Black has just taken a piece on e7.
What are the geometrical patterns here?
  • Black knight is hanging.
  • White queen is haning.
  • Black threathens mate at b2
Nothing special, just normal considerations.
What are the themes here?
  • Pawn promotion by white.
  • Backrank problems for black.
  • Unsafe king for black.
  • Pawn promotion by black.
  • Backrank problems by white.
  • Unsafe king for white.
White has the initiative, which makes all the difference. The first move is not hard to find:
1. dxe7+ Kg8

White to move.
Two new geometrical patterns show themselves.
  • Mate with Bc4+, Kh7 and Qh5#
  • Mate with Bxf6, Qxg7#
Trial&error shows that these patterns don't work yet.
But it reveals that Rc8 is close to being overworked. Can we find a move that accomplish the following:
  • Keep the existing threats.
  • Step up the pressure on Rc8 so it gets overworked.
  • Save the white queen.
2.Qg6          accomplishes exactly that.
Checking possible countermoves show that none of them is worth to worry about:
Be8 and Re8 are met by Bc4+
Rh6 is met by e8Q
Blacks best bet is
2. ... Qc6    covering e8, c4 and keeping an eye on f6.

White to move.
The black queen is overworked.
At this point trial and error must reveal you the final killing pattern. Without that geometrical pattern, you never can find the following move, since it involves a bishop sacrifice.

3.Bxf6   a2+
4.Ka1   Qxf6
5.e8Q+ Bxe8
6.Rxe8+ Rxe8

Black to move. But it is completely over.
7. ... Qf8
8.Bc4+ Kh7

This example gives a good impression of what you can expect from training that is aiming at pattern recognition. There were certainly a few patterns to be recognized, but none of them was uncommon. Maybe the final mate is hard to see if you are not familiar with it, but it was certainly in my database.
This means that for me, more geometrical patterns are of no help in this case. And that is my experience time and again with 2000+ rated problems.

The real problem is making the right considerations. It is easy to astray in that area. Only training that targets exactly this will be of help. One would expect it is no big deal time and again you will find the same considerations. Currently I do a themed problemset with only mates. There are a few considerations that are most common for finding the moves that lead to mate:
  • prevent the king from escaping.
  • look for an invasion square.
  • removal of the guard.
  • add pieces to the attack with tempo.
  • when forced to play a quiet move, look out for counter attacks.
When you have no idea what to move, the move will probably serve one of those themes.
The problem is that making these considerations by conscious thinking is tedious and error prone. Training must accomplish that these themes popup automaticly.


  1. This time it was easy for me. I found the solution pretty quick.

    How comes? Am I suddenly at master strength?!

    Well, here is my secret:
    You posted this puzzle about a week ago? To be exact: it was your post from the 24th of July.

    I remembered the puzzle because I analysed it and discussed it. that makes a problem stay a bit longer in your memory. In a month or so I would not have been able to remember it, but it is just a bit more than a week ago.
    And now I looked at it again. So it will stay with me even longer.

    For the e&t approach: I doubt this can be improved. It is the part that is not possible to be trained.
    We cant change our thinking speed, because our "hardware" will be the same. Only the software can be improved: patterns, guidlines, rules by thumb, reasoning... all this is increased "knowledge" and can be stored in the database "long term memory".
    But trial and error at lighting speed? I need a better brain for that.

  2. Jee, I even forgot that I already posted this one. That's bad.

  3. @Munich,
    For the e&t approach: I doubt this can be improved. It is the part that is not possible to be trained.

    Yet that's the problem we have to solve.

  4. the "thinking speed" can be improved ( i think )->
    Some background

    Trainings example:

    These exercises for mental speed are close related to the board vision exercises of Fritz and to high speed easy tactics exercises. So i think it is possible to improve the e&t approach.

  5. Aox,

    now that you mention it: I forgot about this.
    Well, yes, e&t is possible to improve.
    However (sorry):
    The improvement does not stay permant.
    In that sense it cant be trained.
    What can be done, though:

    We need to solve a couple of very easy tactics as a "warm up" training.
    That gets our e&t on a quicker level.
    So whenever we start CT Blitz tactics, or whenever we start a real OTB tournament, we should have a quick warm up with easy tactics.

    How long does the improved e&t last? Probably as long as we keep doing CT Blitz puzzles, or as long as the tournament game last - until we stand up to go to the toilet or our opponent starts to get into thinking for 20 minutes or so and we get bored and stand up and make a little conversation with others in the club.