Thursday, June 08, 2023

What we know so far

 The past 18 years quite a few facts have presented themselves. All facts need to be met by any theory about chess improvement. What do we know?

  • Papa Polgar proved that talent is not needed when the right education is provided
  • Ton Sijbrands was a champion in checkers, but a mediocre chess player. So it is not a brain thing. He learned chess at a much later age than checkers. So a brain change by aging might be involved.
  • When you learn a Spanish word, you are inclined to forget the same word in German. So the age thing might be a replacement thing of a full brain.
  • From Troyis we learned that you can learn to manipulate knight moves in the brain in a grandmasterly manner. When you plateau in Troyis you can go in two different directions: exercise to make your current skills better or develop a new strategy and make that a new skill.
  • Susan Polgar's simul exhibition showed that it is a skill trick. There is no thinking involved.
  • Mr. Z learned us that visualization must be guided by a logical narrative.
  • Bent Larsen showed me that the vulture sees nothing when it is not guided by system 2.
What kind of skills are we talking about? It is mainly about SEEing salient cues. System 2 does his logic thing and asks himself "how do I do that" and system 1 retrieves some stuff that might or might not be relevant.

How is the vulture educated? Monkey see, monkey do.

Diagram 1. Black to move


Diagram 2. White to move

Diagram 3. Black to move


Diagram 4. White to move


What do these positions have in common? It are all piece sacs where the king must abandon the invasion square, which is then invaded by the queen. Do you notice how totally different the positions are?


  1. The task division is: system 2 asks himself "what should I do?". Then system 1 swings into action and presents a proposition. If it is adequate then system 2 accepts it, if it is not, it is rejected.

    Then system 2 decides on what to do and asks himself "how should I do that?". Then system 1 swings into action and presents a proposition. If it is adequate then system 2 accepts it, if it is not, it is rejected.

    There are a few possible holes: system 1 is not well educated and is not able to retrieve the right propositions OR system 2 doesn't decide between suitable and unsuitable actions and accepts the first one that is presented.

    If one of these holes present themselves, we usually shift to trial and error mode.

  2. Where is the knowledge in the four diagrams? There is not much knowledge involved. After doing these problems a few times, system 1 starts to notice similarities between the positions. If you ask system 2 about it, it starts to produce some gibberish like "It are all piece sacs where the king must abandon the invasion square, which is then invaded by the queen." Nice, but totally useless.

  3. The comparison between ChatGPT and system 1 imposes itself: if the vulture is not educated, it sounds like a hallucinating ChatGPT.

  4. I like your abstraction of the “essence” of the four problems. That nicely summarizes the salient “clues” needed to begin solving them.

    I easily “see” the similarity between Problem 1 and Problem 2. The king is the only piece preventing the enemy queen from getting “up close and personal.” A sacrifice forces the king to move away from the queen’s attacking square. An interesting aspect is that when the queen DOES move to that attacking square, it covers the open “wall” of the box surrounding the king. Although the attacked side can interpose to block the queen check, it does not matter because whether the queen cuts off all the escape squares or the queen in combination with an enemy piece cuts off the escape square is irrelevant. It’s also interesting that a pawn delivers the final kill shot.

    Problem 3 begins to diverge from the pattern slightly in one of the two main variations. If the king captures the knight instead of the rook, then checkmate immediately follows. By a queen move However, if the rook is captured, again, the queen closes the open “wall” of the box. After the knight interposes, the pawn assassin again delivers the kill shot.

    Problem 4 begins to move toward the outer edge of this category. The pawn forces the king away from the focal point, allowing the queen to close the “wall” - except for one escape square. The mating pattern is a Swallow’s Tail or Dovetail Mate, which did not occur in the other three problems.

    It does NOT matter what names we attach to the various themes/devices nor to what category (or categories) we assign a leitmotif for a problem. All that matters is “seeing” (recognizing) salient “clues” and then following them toward a KNOWN mate pattern [knowledge]. THAT is the skill that must be trained and practiced until it becomes ingrained in System 1. The “tricky” part is making sure that System 2 triggers the appropriate question(s) to get System 1 aimed in the correct direction, and also follows up by determining that System 1 is not proposing a simpler alternative that does not really apply.

    1. Most required knowledge is already known. Like pin, duplo attack, clearance, luring away et cetera. But the knowledge is not absorbed as skill. The task at hand is to absorb the knowledge in the form of salient cues.

      What is the repetition for? Not so much to ingrain the matter, but to peel the position like an onion. The first three problems were from a themed set "pawn mate in 3". The fourth was from a set "dovetail mate in 3". I'm sure that we will find all sorts of this invasion in every mate in 3 set. You will find new themes with every repetition.

      In the end is it about SEEing how your pieces work together to accomplish a common goal. In this specific manoeuvre: lure the king away from the invasion square, invade with the queen which forms a few walls of the killbox and deliver mate. The amount of themes is definitely finite.

      The logic involved is certainly not highbrowed, but system 2 must learn to apply it automatically with the assistance of system 1.

  5. Why 3 movers? The art is to keep the effect of 3 moves at the same time in your short term memory. You will see that in the positions that you haven't mastered, you see these effects as 3 separate moves. When you see one move at the time, you can't be aware of the cooperation of the pieces. 2 movers are most of the time too easy, while more movers tend to be too complex. It is about SEEing the cumulative effect of the 3 moves at the same time. So you know what you are doing.

  6. Troyis showed that there are 2 directions to go. Direction 1: hone your current skills. Direction 2: develop new strategies and turn them into skills.

    The same applies for chess. Direction 1 can make you a grandmaster. That is why direction 2 gets hardly any attention.

    When I work on direction 1 in chess, it is just a matter of grinding on. As long as I can gather adequate problem sets, it's just a matter of doing. But
    all work and no play makes Jack a dull boy. So I can't help myself to think (and write) every now and about direction 2.

    That make this blog difficult to understand of course, since both approaches are totally different. Further developing the PoPLoAFun system in conjunction with the Art of Attack in Chess leads to conclusions that are not immediately applicable to direction 1. I hope I don't confuse you too much with that.