Saturday, December 10, 2016

Genesis of knowledge

Due to my previous actions, I have a database of 110 problems which I failed to solve correctly. I already have been putting a vast amount of effort in the analysis of these problems, so it seems logical to use them for the future genesis of knowledge. If I use problems I showed you before, please bear with me. The position might already be familiar, but the yet to derive knowledge is not.

The sheer amount of different "techniques" in the dataset is simply baffling. There are so much different things to know, that it seems almost impossible to distill some knowledge that is commonly usable.

I decided to start with the pin, since that is what the book of Weteschnik does. Let's describe a few things we know about pins. I reckon the pin to the family of the duplo attacks. With one move, you attack two targets at the same time. The head and the tail of the pin. The opponent needs two tempi to get both the head and the tail out of harms way. Unless he has a dual function move to his disposal, he is in trouble.

Already some vagueness starts to cloud my head, am I not skewering the tail piece? What is actually the difference between a pin and a skewer? Or between a skewer and a röntgen attack, for that matter? It seems to have to do with the relative value of the head and the tail piece. If the head piece is more expensive than the tail piece, then we have a skewer, otherwise we have a pin. So what if both head and tail are of equal value?

What emerges is the first clothes hook of our coat rack:
  • Value
The relative values of the two targets, the value of the attacker and the way the targets are protected lead to different interpretations of what you have here on the board, and how the logic way to proceed should be. Maybe "protection" or "defenders" should be the second hook.
  • Value
  • Defenders
What immediately jumps into the eye, is the fact that these two hooks of the coat rack aren't limited to a pin. They play a role in every tactical combination. I think it is good to elaborate on those hooks nevertheless. Later we will find out what is pin specific.

An important factor of a pin is that the two targets are standing on the same line as the attacker. The same is true for a skewer. Some times, the opponent can turn the tables, and change the head and tail of a pin into a discovered attack against your attacker. That can happen if the head piece of your opponent can attack a piece of higher value than his tail piece. So two other hooks emerge:
  • Value
  • Defenders
  • Geometry
  • Initiative
The geometric motif so far has been rather theoretical and of little practical value to me. Yet for a pin it is of paramount importance. For three out of four types of duplo attacks are based on having target(s) and attacker(s) on the same line. And even a subset of double attacks has the same property. We must find out where to look for, concretely.

We already talked a lot about the initiative. Yet describing some practical knowledge has proven to be illusive, so far.

Another subject that comes to mind is the placement of the head and the tail piece. How to put them into the line of attack by preliminary moves like exchanges and such.
  • Value
  • Defenders
  • Geometry
  • Initiative
  • Target placement
An important issue with pins is the temporary immobility of the head piece. How to exploit that?
A piece placement with both targets and attackers on the same line can look harmless at first sight, but a forceful opening of the line of attack can all of a sudden uncork the attack. So:
  • Value
  • Defenders
  • Geometry
  • Initiative
  • Target placement
  • Exploiting immobility
  • Opening of the line of attack
The interesting thing is, that the elements that the different pins have in common, don't differ from other tactical combinations. This means that if we find here some practical knowledge, it will be useful of all types of tactical positions. I didn't see that coming.

We have an impressive coat rack already.


14 comments:

  1. I think pin motif (concept) is much more important then we probably see it. But in the broadest sense as limitation the pieces power or winning material (in a drastic form - mating the King).

    I have solved about 80-100 tactical positions from Lichess and most of my failures were due to not noticing (understanding) the idea of pin/unpin and opening (closing) the files.

    I have a few ideas how to test my hypothesis and I want to see if my conclusions would have any impact on my tactics (in both forms - better and deeper understanding the tactics while playing and in the tactical mode at Lichess or CT).

    I am really frustrated by my totally stupid mistakes. Just 5-8% of tactcial positions (below 2200) are completely new to me and my level of soundness is very low at such.


    BTW. I would like to have a look at your database of 110 problems which you failed to solve correctly. Is it possible to paste these positions? (or share a link to download it). Thanks in advance!

    ReplyDelete
    Replies
    1. It is very important to realize how silly and unnecessary our mistakes are. Only then we are on the same wavelength, and can you understand what I try to accomplish. Don't get frustrated, and be happy that it are just silly mistakes. If it were serious mistakes, it would be much harder to correct our problem.

      You can find links to my dataset here

      Delete
    2. Thank you very much for the set of positions. After solving just a few of these... I came up on the idea what consitutes a difficult puzzle. I am going to write down my observations and considerations and share with you.

      Nowadays I am in the process of solving 1080 chess puzzles from Slavin's workbook vol.7 (book of chess problems/puzzles). I have already confirmed some of Aox and your findings.

      One funny example from my present practice of solving: I glanced at the position for about 2-3 seconds... and I knew the solution (mate in 6). I hope it may be interesting case for you and Aox as I remember you always pointed out the approach (desire) "Even if I solve the more complex puzzles, it is too slow - If I could do it 4-5 times faster it would be a success".

      Do you think solving #6 puzzle in about 3 seconds is success or not? Support your opinion.

      Delete
  2. you try to understand pins.. but what type of understanding will really help you to get better in tactics? The ability to describe / categorize pins will not necessarily increase your ability to find and make use of pins.
    In my eyes the are just 2 things to know about pins:
    "how to find (potential) pins" and "how to make use of pins"
    =~the weakness(es) and the move(s)
    =~vision and calculation

    ReplyDelete
    Replies
    1. Yes Aox, you are right. Bare pin explanation and categorization may not make any improvement. However it may be a very good tool to make a progress at this area. Take notice the pin is a special case of exploiting some weaknesses due to immobility to defend from the attacking army.

      1. Finding potential pins is relatively easy.
      2. Making use of pins is very difficult as you have to spot, use and evaluate:
      a) the weakness(es) and the move(s) [or rather the position]
      b) vision and calculation - especially important when there are so called "illusions" of many good moves
      c) ability to find refutations and choose the best variation among many similiar (good enough for the first sight).

      In case of mating the King (with the use of pins) you have to visualise which squares (and pieces) are attacked and defended. It is especially difficult when you have to visualise and calculate long variation (not to mention with quiet move included). The database of mating patterns (and simple #2 or #3 actions) are a must to do it in an efficient way. At least it is my observation from solving Slavin's (vol.7) puzzles - many of these related to mating the King with the use of pins (and also other motifs included).

      Delete
  3. @ Temoschlucker:

    Would you please elaborate on your own distinction (if any) between "knowledge" and "skill"?

    I ask only so that I can (hopefully!) avoid inadvertently derailing the discussion away from the desired goal(s).

    I personally think skill depends on knowledge (to some extent) but knowledge does not automatically translate into skill. For example, I KNOW (have knowledge of) the mechanical definition of a pin, but I do not have the SKILL to "see" all pins and the ramifications of those pins in all positions. (Substituting "tactics" for "pins" would accurately describe my lack of SKILL.)

    Thanks!

    ReplyDelete
    Replies
    1. We are in uncharted territory here. We tend to treat terms like knowledge and skill somewhat rigid. But how the one transforms into the other, we don't know. Thinking is slow. Knowledge can pop up lightning fast. Is the latter a skill, for that reason?

      The biggest problem I encounter, is that it takes a lot of time for knowledge to pop up. But when it pops up, it pops up lightning fast. This means that the knowledge is already there, stored in the deeper layers of the mind. In cognitive science,cues are supposed to retrieve the knowledge. Like an index to a database, as it were. The more index keys a record has, the more ways there are to retrieve it. So we must build cues from different angles, in order to make it easier to retrieve the knowledge.

      We build cues by intensive study of the same subject from different angles of attack. That's what the theory says. But what that means in practice, we have to find out.

      Delete
  4. "We build cues by intensive study of the same subject from different angles of attack."

    I think you have captured the essence of the training question with this theory statement.

    It is an interesting analogy, comparing human memory retrieval to a computer database.

    Is it also the case in human knowledge acquisition that the more cues (indices) that are created, the slower the process of adding new information or deleting old information, i.e. the longer the required training period for associating cues and knowledge?

    That's the inevitable trade-off in computer databases. I know as a former database designer that I had to consider the frequency of retrieval of data relative to the changes in data (addition/deletion) in order to optimize the database design for its intended primary usage.

    Given multiple cues, it would seem obvious that the knowledge would likely be more retrievable. (A small caveat: if the specific position does NOT trigger any of the stored cues, then the knowledge remains inaccessible.) The hard (unknown?) part would be the training process(es) needed to hammer in those cues that would reliably enable knowledge retrieval. We don't seem to have conscious control over the connection between any specific cue(s) and the desired knowledge. Consequently, we don't seem to have known training methods that will reliably grind the appropriate cues and associated knowledge into our brains. MDLM's method apparently was not repeatable by others. (No need to hash again the question as to whether his method actually worked for him or not; I'll concede that it did not, without argument as to whether he "cheated.")

    I really HATE the idea that we are doomed to blindly wander around in the dark, hoping that "something" "somehow" "sometime" will create an accessible and reliable cue to the tactical knowledge we want to readily retrieve.

    ReplyDelete
    Replies
    1. "I really HATE the idea that we are doomed to blindly wander around in the dark, hoping that "something" "somehow" "sometime" will create an accessible and reliable cue to the tactical knowledge we want to readily retrieve."

      We are not totally blind. My impressive list of methods that don't work narrows the field of investigation considerably. Usually after wandering around in my mind for a minute or more, a cue is triggered in almost any chess problem. We must have a closer look at that very cue that works in the end, and ask ourselves what we need to know to see it faster. The best way to work on cues is to simplify matters.

      Remember that all hooks are applicable to the majority of tactical themes and not just limited to the pin. Everything we can hang on the coat rack will have a broad application. The coat rack is a metaphor for our cue system.

      I'm working on a post about geometry, with a position we have already seen. Let's see what we can discover about cues.

      Delete
    2. Robert Coble said :"I really HATE the idea that we are doomed to blindly wander around in the dark, hoping that "something" "somehow" "sometime" will create an accessible and reliable cue to the tactical knowledge we want to readily retrieve."

      Temposchlucker said : "Usually after wandering around in my mind for a minute or more, a cue is triggered in almost any chess problem."

      The knowledge which is necessary to solve tactical problems is quite simple, you need to know the rules, the value of the pieces and a very few other things like typical draw situations asf. Virtually everyone can solve every tactical puzzle if the visualization skills are sufficient and the person is calculating long enough. Doubleing the calculationtime will increase the the (untimed) performance in solving tactical puzzle by 100-200 elopoints.

      Temposchlucker said: "We must have a closer look at that very cue that works in the end, and ask ourselves what we need to know to see it faster."

      Thats it! I would add, we need to falsify! the false triggers quick too.

      GM Smirnov suggest to implement a system of thinking which becomes automated and then.. after a while..runs unconcious. This method supose to create skills. New aquired knowledge should be implemented in this system of thinking and become this way, by repetition for a period of time" skill. But.. i have problems to do so ;)

      Delete
  5. https://en.wikipedia.org/wiki/Knowledge,_Skills,_and_Abilities :
    Skill - Is an observable competence to perform a learned psychomotor act.
    Knowledge - Is a body of information applied directly to the performance of a function.

    To know the principles of the opening is one thing, to play according to these principles in a bulletgame ( without much thinking ) is a different one.
    Skills can be performed parallel ( driving a car + smoking ) while using knowledge is a sequential thing.

    ReplyDelete