Saturday, September 17, 2011

No concepts, no visualization

The past few weeks Mr Z and I have been discussing visualization in chess. We arrived at the following conclusions:

No concepts, no visualization.
If you have not available the concept of a crocplevier you cannot recognize it, even if it bites you from behind. If you doesn't have a concept of a car, you cannot imagine it. You can't visualize it before the minds eye. If you DO have a concept though, you can't stop the mind from being able to visualize it. That clarifies why blindfold players declare to see concepts and idea's. Not pieces and board. That's why you can play blindfold chess against a much lower rated opponent so easy. Due to his mistakes it is easy to have a conceptual idea of the course of the game. That gives your mind grip.

No difference between high level and low level concepts.
By the mind, concepts are treated equally. No matter if it is a subtle concept how to keep h7, g7 and f7 under control by g4 when a transition to the endgame is near or a more straightforward idea as to put your rooks on an open file or a pawnfork.

No complexity.
The human mind isn't capable to handle complexity. Complexity usually consists of a vast amount of simple concepts. If a grandmaster tells you what he thinks it usually doesn't sound like rocketscience. You just slap your head and say "of course!". If you try it yourself behind the board you find that you lack the concepts.

You have to obtain a vast database with concepts. Who should be the creator of those concepts? Not you! It simple takes way too much time to invent everything on your own. Belief me, after 10 years of studying chess improvement I consider myself to be an expert on wasting of time! No DIY!!
This means that your concepts must be obtained from an outer source. Be it a coach or an author or a chess DVD. For tactical concepts with finite solutions you can use a chess engine. This means: don't waste time by looking for the answer to a problem! After solving 100K+ problems at every regimen you can think of I can state with entire certainty and with no doubt: don't waste time with inventing the wheel on your own. This blog is meant to be a landmark to prevent others from wasting their time. Let Houdini or Rybka do the work. The only thing you have to do is to pour the information into nice little reusable concepts.

Chunks are concepts.
A concept always has some kind of generalization in it. It binds together a whole host of positions and moves into a chunk.

Concepts in chess.
What are they? Concepts describe the relationship between pieces and squares by means of little naratives. Let me give an example.

Black to move.
It is a pretty high rated problem (2046), due to the many possible moves of the king.
Yet it is very easy once you start formulating concepts like:
  • When I move the king it must attack something if possible
  • I must not interfere with the communications of the rooks
  • I must not put my king on an open file that my rooks may need later
  • I must not give a chance for a skewer.
What you see here is the power of concepts. They are not only useful for the position at hand, but they will be useful for a lot of similar positions in the future too.
You see how simple these concepts are. When you have acquired these concepts, it is easy to visualize them. Without them, it is impossible. The concepts guide your visualization. If you put too much time and energy in finding the solution yourself, you will lack the time and energy for summarizing the solution in a narrative. Which is my greatest fail the past 10 years.

Difference between tactical and positional concepts.
Tactics take a very special place in the realm of concepts. Chess is a game of practically infinite posibilities. Concepts are a means to handle such infinite amount by binding groups of positions together. The glue is generalization. Positional considerations are used when there is no finite solution to the current position. When there is a forced continuation, the infinite amount of possibilities is restricted by force. The amount has become finite. Tactical concepts handle forced continuations.

It is evident that the ability to visualize long and complex tactics into quiescence can give you an edge. In fact it is the only ability I consider to be a possible supertrick.
Since concepts are the only means of visualization, a database with tactical concepts is necessary. There is another kind of concepts that plays a role. These are the limiting concepts. For instance: when you consider yourself worse and you have a sacrifice which leads to eternal check: play it without any further calculation about what might happen after that position. In fact there are a lot of these limithing concepts which are very useful in practical play.

In preparation for the Tata Steel chess tournament 2012 I will build my concept-database.
Positionally by looking at chess DVD's about the middlegame for hours on end.
Tactically by looking at the answer of high rated problems at Chess Tempo (what's in a name?) and formulate tactical concepts like the ones above.


  1. You are changing your training often. If you will have any success, how do you know wich training-method is responisble?
    Maybe the Heisman-Quick-Tactics-method did improve your tactical view by 50 Elo-Points, and your new training does nothing and your next turnaments are successfull.. You would think your new method did bring you these points.

    After all what you did try.. the best trainingsmethod should be already found,.. but not recognised

  2. You sound a bit worried.
    The speed-method works and I will continue it one day.That's why I write it down in a blog. It doesn't adress my mainproblem in the core, though.

    After all what you did try.. the best trainingsmethod should be already found,.. but not recognised.

    I have no spikes in my rating that are not accounted for.

    After every few weeks I find another reasoning. Hence a new trainingmethod. But after 10 years I'm more and more spiralling towards a certain goal. Hence the cycli become shorter.

    It is as if I have all moves of a combination and know all tactical elements that are involved, but I have difficulty in working out the move order.

    I don't think this spiralling is an infinite process. I even said to myself for the first time in 10 years: "if this new method doesn't work I'm ready to admit that chess improvement isn't as easy as I thought."

    It's not the first time. It took me 25 years to find out how to fish. After I found the method I gave up fishing since it wasn't interesting anymore, btw:)

  3. According to your Rating Progress of the Knights Errant, you improved by 324 points to 1856 as a result of solving problems. The average rating in England is 134 ECF or 1724 ELO, so 1856 is a little above average in England, and probably also in your country. Apart from MDLM himself, you are the only player in your table who has left behind a record of what he did, and improved by a large amount to a level above that of the average player. That is a huge achievement. 1x100K problems to improve by 324 points is probably about par for the course (actually your link says that you improved by 170 points by solving 11,000 problems, which looks amazing). 5x10K problems might be as good, 10x10K might be better. MDLM’s 7x1K is a no hoper, unless you are very weak at the outset. After this training you seem to have tried just about everything else. How many rating points have you improved as a result?

  4. Here and here you can see a more exact description between my experiments and rating. From my peak I'm now about 60 points lower. Probably due to lots of experiments and little consequent serious training.

  5. BK,
    now I read your comment better I see you probably found those posts.
    Yeah, that 170 points in 11,000 really felt good and convinced me that much more is possible. I was in the flow by then. I have been searching for the exact constraints of training ever since.

  6. I meant 10x5K rather than 5x10K. nx5k is roughly where I am now. My highly dubious calculations suggest that I have probably improved by about 300 points (most likely to about 2000, if I brush up the rest of my game). I have already learned something close to a majority of the new 5 move combinations that I encounter. After 10x10K I should have learned something close to most of them. A typical 1800 will have to work out most of these from scratch, so I should have a big advantage over him. However, typically he will have learned chess as a kid, read dozens of books and have a reasonably sensible training programme. A lot of focused work is needed to pull away from him, and I always found winning my share of points against 2200+ players tougher. Chess is not an easy game!

  7. The idea of concepts is already an old one.

    That is why people who do know always say that when one studies tactical puzzles that it isn't the solving part that brings you more strenght but knowing the position of the pieces and the concept behind that position that brings you more strenght.

    So good luck with your new way of training. Hopefully it will bring you many rating points.

  8. @BK
    The fun part is that I made my biggest progress without any repetition:)
    Thanks for the cheering!

  9. You are still repeating the underlying patterns, you will still learn them eventually. If you are strong on pattern knowledge and weak at applying them, once through may be better. Why give up on what worked for you? Given your 60 point slippage, I expect that you could gain 100-200 points just by going through your old problem sets; and the next time through, you should still gain some points, but not as many…

  10. "The fun part is that I made my biggest progress without any repetition:)"

    Without ANY repetition of ANY type -> No memorisation = No learning = No improvement