Saturday, October 29, 2011

Difference between patterns and concepts

A concept is something that is totally different from a pattern. And I mean totally. A concept is created by the mind in a conscious way while a pattern is assimilated unconsciously.

In a chessgame there are way too many facts to handle for the mind. By summarizing facts into concepts, you are able to summarize thousands or even millions of similar positions into one single concept.

The weakness of pattern recognition is that it is too vulnerable for details. One pawn on a3 in stead of a2 can change the whole character of a position. That is true for every pattern. That makes it difficult to find the truth about a position by means of patterns.

Concepts are a means to simplify complex positions. It is our way to cope with a complex world. Inevitably you will loose details when simplifying matters. But that outweights the fact that it makes the game more managable. By far.

Now what are the consequences of this statement? That is easy to miss.

First of all it means that the method I was working with can't be the right one. I made an attempt to free up the strain on my STM by transferring tasks from STM to LTM. And allthough I found a method that indeed accomplishes this, this is not the final solution to the problem. After a hard and long journey I reached the end of the rainbow, but it was the wrong end again.

With concepts you summarize facts. In stead of shuffling around multiple facts in your head at lightning speed, you shuffle around a few concepts at an easy pace. Thus simulating speed for the spectators who think you are still shuffling around a vast amount of facts. The strain on the STM is lowered by summarization.

This all might seem very arbritrary but it has a huge impact on the way to study. When solving problems the emphasis is on finding the solution. With conceptbuilding you start with the solution and build from there. From this point of view, solving the problem yourself is in fact a waste of time. Especially when you solve problems under time constraints, when there is no time for evaluating the solution.

As a novice to chess, we started with very high level concepts and general rules. The longer we played, the more low level the concepts became and the more specific the rules. But somewhere in that process we became hypnotized by moves and variations. And look at us now. We are so addicted to moves and variations that we can't walk along a chesspositon without looking at the board and working out moves and variations. When that happens, we are no longer adding concepts to our repertoire.

When you evaluate a solution, you try to generalize it. When you manage to do that, you have build a concept of a generic solution. The concept is no longer applicable to the problem at hand alone, but it can be used to solve thousands of similar problems. The task of writing down a set of concepts has a lot of simularities with the Stoyko exercise. It is pretty daunting and out of your confort zone, so it is no wonder we have fallen for the easy path of contriving moves and variations instead

You can test easily if you have done your job well when building a concept. You can only visualize something of which you have a concept. So when you are able to visualize the whole solution with all its thickets you are ready to move on.

Since visualization and tactics have such a close relation, we can postulate the hypothesis that conceptbuilding will have a huge impact on tactics.

@Munich: The B-method has nothing to do with this.


  1. An interesting thread!

    I certainly agree with the idea that in general we see something only, if we have a concept for it.
    Though until now, I haven't seen it consciously applied to chess learning like you describe it in your recent posts.

    I can easily imagine how e.g. the notion of good/bad bishops can be framed into a concept, etc.
    I'm less certain if tactics can be conceptualized in a similar fashion.

    Overall, I'm rather optimistic that it is possible to mold a lot of chess knowledge into concepts.
    However, I don't know if this really is a (fast) track to chess improvement. Just because you have internalized a lot of concepts doesn't necessarily mean that this really translates to superior performance at the board.

    For the sake of the argument
    let's assume we have framed the relevant stuff on good/bad bishops into a concept. So far so good.

    We want this concept to be productive -- so we want it to pop out on its own whenever it is needed, i.e. whenever the position on the board does have the relevant characteristics.

    I guess it would pop out, if it's the only concept we have or if it's one concept of a small number.
    However, in actual fact we would need a rather large number of concepts ... and then I wouldn't be as sure if concepts really pop out on their own.

    In another post you wrote:
    "If you DO have a concept though, you can't stop the mind from being able to visualize it."
    I guess that is the step I don't follow until now. I quite often did miss something about good/bad bishops at the board
    although I certainly do know quite a few things about this matter.
    (So maybe knowing quite a few things and having a concept is not the same thing, no?)

    So why are you convinced that you cannot stop the mind to see the concepts, once you have them?

    When all is said and done,
    I guess the question will be ...
    if it works.
    What are your impressions so far?

    I really do want you to succeed, so don't let these speculations deter you from your path.

  2. You wrote: "With conceptbuilding you start with the solution and build from there"

    With that I very much agree. If I think about my training then I remember, that I commented very often that the puzzle "is a fork" or "is not a fork", but a discovered attack.... and so on.
    My intention was to get a better quality into the tags. However, thinking about your sentence, I probably put in pretty much time into finding the concept in the given position. Because besides of the intended tag to train, I also voted up or down other tag-elements of the puzzle. In some variations the solution is a discovery, in others it is a fork, so I tagged both. It could be possible, that this effort is part of my "success" (I still need to show more of it at CT, but besides of this detail, I know I became much stronger, beating occasionally people in the chessclub I havent had a chance before).

    Starting from the solution: When I do training with my little son, I give him lots of hints --> getting his attention to the "concept" of the position. Though he hardly does train, my impression is, that this approached in these rare occasions allready improved his chessability more than if he had to find the solution all by himself.

    I also need to admit, that I havent understood everything you have written, though.
    Deliberate praxis (trainig puzzles you did wrong or you did slow) is speeding up these "trouble puzzles", too. The speed of doing puzzles is not the skill itself, but a side effect. You know the concept well - you will become faster. The time I needed to solve a puzzle tells me about my knowledge of the position. So speed is a good gauge to find out your weakness. Also: if you take too long for a puzzle - how many puzzle are you able to cover per day. Doing 10 puzzles a day wont do the trick.
    On the other hand - if you do puzzles that you are able to do within (let's say) 10 seconds, then these puzzles are hardly worth to be done. You need to find easy puzzles, that are easy for others, but not for you. richard (owner of CT) was so nice to give me a special filter: "use most recent time". With this I can filter for puzzles where I needed more then 10 seconds last time I approached it. In that way I do perfectly deliberate practice.
    regards, Munich

  3. @All,

    It certainly pays off to take the time to explain my thoughts more clear. It leads to inspirational comments which, after all, is my main goal for maintaining this blog.

    The popping up of the concept and the visualization of it are two different matters. What is necessary for popping up is usually known as association. The more cues, the greater the chance to popup. I haven't investigated this in this new context yet, but I will.
    A thought process should help her

    Once a concept has popped up though, the visualization is possible. Which means that you can use it in a calculation.

    The concept of the good, the bad and the active bishop is a somewhat higher level concept than a concept like "my piece is attacked, let me put it on a square where it doesn't stand in the way of my other pieces".

    My first impression is that it does what it says on the tin. But I'm biased.

  4. @Munich,

    Of course it is a good idea to take speed as a gauge.

    if you take too long for a puzzle - how many puzzle are you able to cover per day. Doing 10 puzzles a day wont do the trick.

    No. It is about "how many positions does this concept cover". Not about how long does it take to formulate the concept. A concept is a generalisation. You look at the solution as a single instance of a more general idea. I do typically one (complex)problem in two hours. Leading to the formulation of 3 to 6 usable concepts. 1 concept can cover millions of positions.

    Remember what Stoyko said. Everytime he did the Stoyko exercises he gained 100 ratingpoints.

  5. Could you give some more examples of concepts as you understand them?
    I'll do my very best to post some inspired comments in return.

  6. Agree with the core notion.


  7. My last post found its way to the next article. It is not easy to find the end of an article in this blog - I keep overshooting!

    The obvious way of training, and that one that most if not all strong players use is:
    * Practice finding moves (e.g. solutions to problems).
    * Find out where and why you went wrong.
    * Try to fix the problem.

    People learn better by trying to find solutions them selves rather than looking up the answer (active versus passive learning). You lose this advantage if you start fro the solution. You also loose feedback on what you are doing wrong.

    Its not easy though! I am doing some slow training right now, and I keep writing down: consider all defences, consider all move orders, find a good move look for a better one, and count the pieces. Getting my brain to do it is not easy!

  8. are some of my comments in your spam folder?

  9. @Aox,
    I changed to a new computer past weekend so they might be lost. Are you able to reconstruct them?

  10. The Stoyko excercises appear to me like going the road "standard puzzles" at CT on "hard" mode. I was able to improve in rating in Standard mode, but I became slower and slower. I needed often more than half an hour to solve such a puzzle. An hour was not seldom.
    So I reached the point, where each puzzle took me a long time. I calculated lots of side variations, and afterwards I often checked my variations with a programm. O.k., I didnt write down my analysis, but I knew what I was thinking anyway, no need to look at a paper to remind me.

    But I feel, that doing very hard complex puzzles did not help me to improve in Chess. First it did help me, though.
    The learning effect of doing a stoyko excersize is probably little different to solving very hard CT Standard puzzles. The main difference is, that a CT puzzle at least has a definite solution, while a stoyko is more on the fine adjustment tuning.
    Doing massive amounts of easy tag-related puzzles helped me much more. I learned typical tactics, and was able to recognise them later in my real games. You had this experience, too: You saw suddenly a lot more forks and pins, after you did fork- and pin- puzzles. Though you did not stick to this method, I sticked with it a bit longer. And yes, you found here a way to improve - but you seem to have forgotten about it. But it really works. I can tell, because I have the advantage of knowing what is going on in my mind. I cant give a prove, but I know I remember CT Puzzles such as "ah, this position is similar to a CT puzzle I did train. Just a pawn is missing. Wait, maybe I can attract an alternative piece to this square? Yes I can!" And suddenly I find a very impressive tactic within a blitz game, and my opponents are astonished about my tactical strength. Actually I am astonished about myself to. This training enabled me to "work on tactics" that are not there but almost. It is as if I would "dream" the critical position before it appears on the board. I know where my piece should stand to give me maximum attacking possibilities. So all this knowledge stored in my LTM improved my positional play. I can know much better judge if a position is favourable or even.

  11. I somehow come back to my thoughts from my first comment ('An interesting thread ...')

    Over the last years I've spent quite some time trying to improve my chess and to some extent I did succeed. (And to some extent I didn't.)

    For me the hard part was not to acquire new knowledge, knew "stuff". In general the hard part was to translate this into chess skill.
    My chess knowledge certainly is way ahead of my actual chess skill. And I guess that is what happens often when you read lots of chess books ...

    You mention cues, association and thought process ...
    Somehow thought processes don't do the trick for me. It seems to be too artificial or too cumbersome during actual play and I did not really bring it to life in any consistent way.
    (Maybe that is because I mostly play rapid chess - maybe.)

    I'm certain this focus on concepts is a good method to understand things thoroughly, it prevents you from just superficially looking at positions, it might prevent you from a 'reading and nodding' attitude. -- And that is a good thing.

    I'm curious to see if this really translates to enhanced chess skill, i.e. performance at the board. (Because this step has been the most difficult for me ...)

    I really want you to succeed.

  12. P.S. The post (Friday, November 4, 2011 11:33:00 PM GMT+01:00) was from me, Munich.