Monday, February 01, 2016

Elaborating on a thought process.

In the pursuit of a universally usable thought process (TP), the following categories can be identified:

  • Control beforehand.
  • Looking for inspiration.
  • Control afterwards.

I use the term "control" in stead of "checks" to avoid confusion with "giving check". I like to keep the TP as minimalistic as possible. It makes no sense to cover every exotic idea you can think of. The goal of a TP must be simplifying a complex position. The idea is, that as much as possible, the TP must be automated. That implicates that steps that can be envisioned as a picture are preferred. I formulate the steps as questions, you need to interrogate the position for answers.

Control beforehand.
  • Is my king in check? 
 In an OTB game, this step can be skipped. Since if you don't know that your king is in check, you might be better off when seeking a hobby that suits you more. Although in time trouble, you see sometimes that kings are checked without both players noticing it.
When solving puzzles, it is a good idea to check if your king is in check, because if it is and you don't know it, you are wasting a lot of time. While controlling if your king is in check might take less than a second. It is a typical board vision exercise.

Looking for inspiration.
When doing tactical exercises, there are two questions to be asked (= my personal TP):
  •  Which are the squares from which to attack?
  • Which pieces are overworked?
When you are solving mate in ones, you don't have to look far for inspiration, since, well, you already know what the position is about. You better ask these questions:
  • Is the box around the king closed?
If the answer is no, the following question should be asked:
  • Which move fixes the hole in the box around the king while giving check at the same time?
If the answer is yes:
  • Which move gives check?
In general, the category "looking for inspiration" should provide a list of candidate moves. For other problems than mate in one, I probably should dive deeper in the matter. But for mate in one, these two questions suffice. Checking the box around the king when doing mate in one "hard" is perfectly suited to train aura vision, a subject that happens to be a major weakness of me.

Control afterwards.
This category adds standard chess knowledge to a move. It takes in consideration all possible logical answers to your move.
  • Is the piece you intend to move pinned against your king?
  • Is the square that you intend to move to protected?
  • Can the opponent interpose the intended check?
  • Creates the intended move a new hole in the box around the king?

If you can answer all the blue questions from the three categories within, say, 5 seconds, there can't be going on much thinking in your head. That would be good enough for me. Since that equals about 12 mates per minute, that would be an outstanding performance. For mate in one "hard".



30 comments:

  1. I think there are several Boardvision exercises related to M1

    Find all Checks.. already existing
    Find all pieces controlling squares around the king / limiting the kings mobility
    Find all forbidden squares around the opponent king
    Find moves to attack 2 (3..) given squares with one! move at the same time

    There is a further execise with Plan B component:
    Name the mate-pattern ( like Dovetail )

    Every single of these task should be improvable

    I think: The/A critical skill will be to see by looking (~only) at the opponent king for just a sec, which squares are his escape squares

    ReplyDelete
  2. From M1 easy to M1 hard, the problems add up. The composed mates are usually in the middle of the board, so there are more squares around the king, there are more temptations of squares were you want to give check but which are protected, and more pieces of you are pinned to the king. M1-e takes me 5 seconds per mate, while MH-h takes me 15 seconds per mate. So the extra instances of the subtasks cost me 10 seconds extra, at average.

    All subtasks are quite common. When I did a few blitz problems at CT today, I noticed that a lot of errors I made were caused by errors in the "control afterwards" category. That means that the move I did was not checked for protected attacking squares, pinned pieces etcetera. So mastering the subtasks of M1 should very well transfer to all tactical problems.

    I don't think all mate patterns in the middle of the board have a name. When mate, you can freely change the box around the black king by interchanging whites controlling pieces and blacks blocking pieces and the rim of the board. It remains mate. There are certainly a few common categories, but studying these are M1 specific. So the effort put in learning these doesn't transfer to other tactical problems. I like to prevent that.

    Studying M1-h might very well lead to an increment in speed at M1-e, since M1-e contains problems that come very close to composed problems, and these will devastate your average M1-e/min.

    ReplyDelete
  3. mate and mate threat are very common in tactics..40 % ?? So a special mate training can make some sense.
    But i think you are right. There is no big difference in the performance of a tactician, if the tactic is easy or hard, many moves or few and what tactical motive ever the timed rating is always almost the same. Someone which will look for XYZ first will be a little bit better in XYZ but thats it.
    So looking for not to specific aspects makes sense.

    You can solve a m1-e 3 times quicker than a m1-h.. what is 3 times more often in m1-h than in m1-e.. possibly squares around the king? Or pieces controlling squares around the king?


    ReplyDelete
  4. Composed mates make a lot of use of "temptations".
    So a lot of checks are on protected squares, or black can interpose the check with a piece.
    Besides that, checking the box around the king for holes takes more time in the middle of the board than at the rim.

    The time I need to check the box for holes already diminishes somewhat. But checking if the check is not on a protected square or whether a black piece can interpose continues to consume an awful lot of time.

    ReplyDelete
  5. The beauty of M1 is that there is no future position. Everything is right in front of you on the board.
    If you don't master this perfect and at lightning speed, you can't expect to do the calculation of future positions well. It really is al about mastering the basics at lightning speed.

    ReplyDelete
  6. M1 dont need any big memory, deep visualization, knowledge about won (endgame-)positions... so it is not "everything"
    In Chess(tactics)
    -there are "fast" things: thats board vision, things to be detected "as a face" or done "by instinct".

    -and there are slow things: like deep calculations, analysis based on concentration, memory of already calculated lines asf.

    -and things based on knowledge or experience

    The first both things are connected, but they are not the same.

    young folks for example dont want to think more than a few sec, they just make a move after say 20sec, or they cant see a position in 2 moves sharp anymore. even if they would have 1 hour to think.. they cant use the time, their k is small. Systematic thinker with good memory and the ability to play blindfolded can use such an hour very effectiv, they have a big k.

    Tactic is the/a laboratory for chess-improvement ( because there are chess tactic servers ) and m1 is the/a laboratory for "fast tactics"

    The Method for m1-improvement is the interesting question. If we cant find a good method to improve decisiv in m1.. then how can we hope to find a method to improve in tactics as a whole and if we cant improve in tactics decisiv.. there is IMO no chance to improvd "decisiv" in chess as a whole.

    On the other hand, to become worldchampion in m1 wont make us beating magnus. But i have the strong feeling that improving in m1 will improve our whole chess.

    ReplyDelete
  7. I agree.

    Although I'm not familiar with the K factor stuff. Is it related to the K factor in mr Elo's calculation?

    ReplyDelete
    Replies
    1. If you think twice as much your performance increase at average K- elopoints.
      K is usually somewhere between 100 and 200 depending on the person and situation.
      Empirical Rabbit did calculate K=200 using the data of Chesstournaments with timehandicap. Using different methods i did averaged K to be somewhere near 140.
      You may see the effect of longer thinking for example here :

      http://chesstempo.com/chess-problems/104160
      Problem Blitz Rating:1360.8
      Blitz Av Seconds:00:28

      Problem Standard Rating:1141.1
      Standard Av Seconds:01:23

      Problem Mixed Rating:1250.6
      Mixed Av Seconds:00:55

      Thinking 28 sec the problem is rated 1360.8 , thinking 55 sec its rated 1250.6 and thinking 83 sec its rated 1141.1

      So it seems! that the "K of this problem" is (1360.8-1141.1)/log2(83/28) =~ 140
      But.. there is no timepreassure at Standard and the level of the standard rating is not exactly the same as the level of the blitz level and, and, and.... So: the calculation above is a simplification.

      A "K" ist often used in Math in the combination with logarithmic or exponetial functions. Empirical Rabbit did give "K" its name.

      Delete
  8. If the elements are working in an excellent way - you can expect an improvement. I do not know if the barrier of 45 is a magical one, but for me it seems so! Even after solving another 1600 puzzles (M1-e) I could not break 45 MPM.

    Working on pins, attacked and defended squares, covering lines, escape squares and "King's box" closing is a KEY to progress at this specific task. However I have some doubts if breaking 50 MPM is possible for most of the players. Simply because it may require fast processing power (of your brain) and/or fast "clicking hand".

    Aox's puzzles (M1-e) are EXCELLENT because they contain every type of checkmate I have ever seen or solved. The only exception is mating with a 0-0 or 0-0-0 move.

    BTW. There are some specific piece configurations around the King that helps to recognize the TYPE of mate. I think it may be a good idea to classify and categorize these typical checkmate finals and patterns. For some of these your processing power (recognition) may increase up to 0,2-0,4sec per (specific) puzzle!

    BTW.2. I am more than curious what score would a typical FM, IM and GM achieve. It is hard to believe that any of them could break 60 MPM (without an extensive training and memorization of the puzzles).

    What do you think about this idea? (classification and categorization)

    ReplyDelete
    Replies
    1. @Tomasz,

      I agree with BTW2.

      As said, I'm not a fan of M1 specific activities that don't transfer to other area's of tactical problems. Yet I might give categorization a try with a one or two posts. I'm not sure. I work with composed mates (M1-h), with mates in the middle of the board, which most of the time have no specific name. So the names must be invented by us.

      I'm curious how you score at M1-h.

      Delete
    2. I don't think that above 40 mates per minute the measurement is very reliable, due to the increase of noise of slow hand speed. At least for me.

      Delete
    3. I believe above 50 MPM this problem REALLY starts to exist. If you have a big touch screen monitor you could do it with you hard (or like some kind of a "electronic gun"). Below 50 MPM the real problem is to be too weak at the necessary (critical) subtasks we have to use at solving #1 (especially easy ones, as I have no idea how high score I could reach when solving #1h).

      Delete
  9. I score 15 seconds per mate at M1-h (4 mates per minute). Today I did the exercise "find all checks" and I scored a measly 12 checks per minute = 6 seconds per check. So improvement with "find all checks" might transfer to M1-h

    When busy with M1-h, I "feel" that looking for the checks and see if the checks can be refuted is very time consuming indeed.

    ReplyDelete
    Replies
    1. To improve in "Find all Checks" will be not enought to make M1 improvable IMO. Even the find all checks might be to hard for you to improve it as a whole. I guesstimate!! that a decision has to be in the sub 2 sec range to be improvable.
      Do you trace from the king backwards or do you scan all pieces? There are related exercises at Lucas :find all moves in the Chess for Zebras section and the attacker and defender exercise at fritz and Chessgym are in this direction too

      Delete
    2. Finding just all the checks eats up 6 seconds of the 15 seconds I need for a M1-h at average. So speeding it up might have an effect.

      It's a bit weird. Some subtasks seem to speed up by a mental effort to visualize them:

      Check of the box around the king for holes.
      Check whether a square is protected.
      Check whether a check can be blocked.

      While other subtasks seem to need a "Troyis-like" approach by just training them in automatic mode, like finding all checks.

      I'm not sure if this observation is right or which cause that would have.

      Delete
    3. If automatic training is sufficient or not is IMO mainly a question about the improvability of the main (dominating) sub task
      1) Check the box around the king consist of a lot! of (complex) subtasks
      2) Check whether a square is protectet is already such a subtask of 1)
      3) Scanning for pieces which attack a given square is a subtask of 2
      4) Answering the question does a given piece attacks a given square is subtask of 3)

      As higher the tasknumber as more "support" is needed to improve it
      and usually a task has several! different subtasks ( and even alternative task-sets? .. "Plan B's")

      It depends on your present skills what is Troys-like ( i use the term improvable ) and what needs further support ( like a thinking process, visualisation..)

      Munich did use the analogy of speedreading: first you read letters.. then whole words..

      Delete
    4. Oops forgot: Find all checks "must" improve m1 but that is IMO not enough to make m1 itself improvable as a whole. It will be still necessary to improve other subskills

      Delete
    5. When I started with "find all checks", I started in "raging madman mode". Just trying to find all checks as fast as possible without further ado. No thinking or evaluation or visualization allowed. Just like I did Troyis in the past. But I already refound my discipline, and I focus on conscious visualization-II in slow tempo again.

      After all, that's what I want to find out. Does the slow method (conscious visualization-II) work or the fast method (unconscious automatic Troyis/raging madman).

      Delete
    6. If your borderline score in find all checks is 12 then find all checks ??might?? be not improvable for you.. ( as i said i guesstimate you need to make a decision in less than 2 sec ), You will have to give it further suport.
      But that you can check.. simply try to improve FAC for a while.. you will experience soon if you can improve this way.
      And then you get back to slow mode ( thinking process-visualisation..) and see if it gets improvable then

      Delete
  10. What average do you guys reach with "find all checks"?

    ReplyDelete
    Replies
    1. Could it be the best score I can get at present? If yes, I can try this FAC for a while and tell you how weak/strong I am at this ;)

      Let me know if you see any progress. At present I cannot break 43.00 MPM even if I try HARD.

      What is the most convincing refutation to the statement: "solving #1e [very fast way] requeires memorization of massive amount of checkmate patterns"?

      Delete
    2. Bad news :(. I barely score 23-24 MPS at FAC. It is clearly visible I have to improve it AT LEAST up to 30-35 MPS to see REAL improvement at #1e.

      Would you be able to write an article at your blog - related to "how to find all checks without missing anyone - REALLY FAST"? I would appreciate it very much!

      BTW> What is your best score at FAC? And what's Aox?

      Delete
  11. I score 12 checks per minute at FAC. But I just started, and I assume it will improve. For now, I focus on FAC.
    It is funny that you faster recognize a mate than a check.
    I will think about the refutation of your statement later.
    I don't know Aox scores at FAC and M1-e.

    From time to time, M1-e confronts you with a M1-h problem. This screws your average. It is a good idea to try M1-h for a short while, so you know where your problems lie. I'm curious to see your score.

    ReplyDelete
  12. My weakness are at least a few ones. I am not going to start M1-h unless I will break 50 MPM at M1-e.

    However I have a GOOD NEWS! :). What is it? ;) Look below!

    Done: 1846 (puzzles) - Check / min: 27.10

    It is a pathetic score in my case (FAC below 30!). Unfortunately I am so sloooow (weak) at finding checks! Anyway I am going to work really hard to realize these goals:
    1) break 30 FAC
    2) break 35 FAC
    3) break 40 FAC

    And after that I will see if it causes the improvement at M1-e. It is a SHAME on me that I barely score close to 30 FAC! It is ridiculous to dream about breaking 50 MPM at M1-e... without obtaining at least 40 FAC!

    BTW> If you want to help me - please write the algorithm of finding all the mates. After I will use it - I expect a SIGNIFICANT improvement at FAC (from 25 to 40!) and after that at M1-e (from 43 to 50!). You will be one of my helpers if you can help me now :). Thanks in advance for everything you are doing: sharing your knowledge, experience, thoughs and testing! :) It is a great fun!

    ReplyDelete
  13. Just before I go to bed. My big weakness is related to "do not take/attack" in a such way. What does it mean? I have ingrained too deep that state of chess mind. I could describe it this way: "Queen cannot takes a protected pawn or piece nor it can goes directly around the King's aura to be taken for free".

    My lack(serious) of improvement is MAINLY related (connected) to this flaw. It is a very serious limitation to my progress at the specific tasks. I can find 90% of the necessary moves, but about 10% of these are "hidden" at the flaw described above and the flaw related to "(very) long moves". If I could eradicate both of these I could score:

    1) about 50-52 at MPM #1-e.
    2) about 30-35 at MPM #1-h.
    3) about 40-45 at FAC.

    These are my estimation. We will see how much effort and what subtasks are needed (especially in my case) to reach these numbers. Maybe after that our friend Aox would change his mind related to "we even cannot make significant improvement at MATE in 1 puzzles".

    ReplyDelete
  14. From the values of Tomasz at Chesstempo i know that Tomasz is twice as fast than i in "easy" CT-blitz puzzles. His initial values at m1 and at FAC are in the same range something like twice as mine but i had a FAC falue of ~18
    So Tomasz should have a CT-Blitz rating of ~~2030-2130. But it was way below that, his k is low ( meaning he is to weak in hard problems / to strong in easy problems ) . Now we see that his FAC value is astonishing "low". IMO this all indicates that the thinking process of Tomasz is not organized enough for reasons unknown.

    An other explanation for his astonishing FAC / m1 ratio is that Tomasz m1 performance is mainly based on mate pattern recognition.. not that this would be impossible but i have doubts here because the relationship between his m1 to my m1 ratios where about the same as in CT-blitz.

    @Tomasz is suggest you try some "Find all moves" at "Chess for zebras" at the free program "Lucas chess". That may help to structure your thinking.

    I think it gets clearer now what the necessitys of adult improvement in vision are and how they are related to improvement in tactical skill.

    At the moment i realy dont feel like writing any programs, too much to do and to few energy :/
    If i write any training progs then maybe a : "find all escape squares" or "find all king restricting pieces"

    ReplyDelete
    Replies
    1. I could write them, but I don't know what is the best game database to extract the FEN list of positions and re-use your scripts.

      Delete
    2. You can take any pgn database in the net ( and merge them? ) or buy one, look for interesting positions, generate the fen and the list of correct answers and store them as list in a txt file, thats it .
      The work is the part with "finding interesting positions" - you need to find criteria for that .

      Delete
  15. [Quote] "Anyway I am going to work really hard to realize these goals: 1) break 30 FAC"

    I have just broken the 30 FAC barrier (just a little bit, but it is a fact). It has been relatively quite easy due to the good initial score. Reaching 40 may be much harder and requires to work out discovering "mate lines" and "mate aura".

    Now I am in the process of getting into 35 FAC. After that I want to achieve 40 FAC.

    ReplyDelete