Tuesday, February 23, 2016

Little disruptions with big effects

After doing the following exercises:
  • Find all checks BLACK K+P ONLY
  • Find all checks BLACK K+WHITE N ONLY
  • Find all checks BLACK K+WHITE Q ONLY
  • Find all checks BLACK KING ONLY
 and improving them to more than 35 CPM, I was quite a bit surprised to notice that I still struggled with FAC No Queens.

All the problem causing pieces like queen and knight are mastered, so why would the simple pieces like rook, bishop and pawn cause any trouble?

Find all checks for black and white
 It is the sheer fact that I must shift my attention from piece to piece, and from black to white, which takes time. Every move has a little detail that disrupt a fluent processing.
  • Rook h4 can take h5 with check but not e4
  • Bishop a3 can give check on d6 but not b2 since it is blocked there
  • Knight d2 has two unequal checks since one of them is a capture
  • Pawn moves like f4+ have a tendency to be overlooked.
The result being a disruption of thought due to tiny details, counting up to a quantum delay causing a factor 2.5 decrease in speed. Or maybe I should say, a disruption by thought, since the tiny details cause many little yet time consuming thoughts.

OTOH, I feel that improving FAC No Queens to 35 cpm should be no problem. Of course FAC as maintask can't be done faster than its slowest subtask. But what will be the result when I have upgraded all subtasks to 35 CPM? Will FAC take off from 35 CPM immediately without having trained it separately the past weeks? Or will the combination of the different subtasks cause mental disruptions in its own right?
Or have we overlooked a critical subtask, and will the overall effect of the training be zero on the speed of the main task? We are going to find out.





20 comments:

  1. FAC is a Board vision of second order, there is a piece, which moves to a square from where it could move to the square where the king is atm. The related Boad vision first oder is "attacker", FAC = 2 times attacker.
    I suggest to start there, improve it to a "gm-level" ( i did stop at 40+ ) and then continue with FAC. At least i had no problem to improve attacker..
    See here : http://www.chessgym.net/usr_ranking_att.php

    ReplyDelete
    Replies
    1. No worries, I just started today with FAC No Queen, and I'm improving. I was only surprised that it was more difficult than I expected. When everything else fails, I go back to exercises of the first order. But so far, I managed to improve at all the exercises of the second order as mentioned in the post.

      Delete
    2. Tempo said:
      "After doing the following exercises:

      Find all checks BLACK K+P ONLY
      Find all checks BLACK K+WHITE N ONLY
      Find all checks BLACK K+WHITE Q ONLY
      Find all checks BLACK KING ONLY

      and improving them to more than 35 CPM, I was quite a bit surprised to notice that I still struggled with FAC No Queens..."

      I have an alternative explanation why you still have problems with FAC-NoQ
      .. You might not have realy improved the other subtasks. The improvement in the score might be based on the memorisation of the few exercises and not be based on improvement in skill
      These sets are extreme small and the set FAC-NoQ is just mutch bigger

      You can test, if this is the case, by mirroring the exercises and check if your performance is precise at the same speed as before

      Replace in the HTML file
      //flipboard(Math.random()>0.5);
      By
      flipboard(true);

      then all exercises will be displayed mirrored left to right, this will fuzzify your memory a little and forces you more to use your skills.

      Delete
    3. LOL. Aox said: I have an alternative explanation why you still have problems with FAC-NoQ

      There is no need for an alternative explanation at the moment. Since there is already an explanation that suffices: I haven't trained this yet. I only started yesterday. Since then, I'm steadily advancing.

      My surprise was about the fact that after improving the problem pieces, the no problem pieces cause a problem. The surprise is not about not being able to improve, since I am able to improve. At least for now.

      So far I don't recognize any position, so memorisation doesn't seem to be an issue just yet. I will store your advice somewhere I can find it though, just in case I need it anywhere in the future.

      Delete
    4. I made some statistics about related experiment with easy mate problems years ago. I made a set of several thousand easy mate in x problems.. The problems where presented random but..i did not improve "at problems i did not see before". Analysing my data i found that i had a higher score at problems i did see less than ~8 days before. I did not had the impression that i did memorise any of these, i solved ~~300 each day.
      8*300=2400.. a set of 2400 of these puzzles and i would have thought i would improve.. but my set was bigger and i did start wondering whats going on...
      You memorise unconciously.

      Your method looks like an attempt to "learn" all FAC-puzzles.. maybe i should find my old prog and increase the number of FAC puzzles by a factor 10

      Delete
  2. I will give a try to these puzzles as you wrote in your post:

    Find all checks BLACK K+P ONLY
    Find all checks BLACK K+WHITE N ONLY
    Find all checks BLACK K+WHITE Q ONLY
    Find all checks BLACK KING ONLY

    And we will see if I can score at least 40 (CPM) at each of these.

    I am really interested how much memorization can influence the results. I can recognize instantly probably even up to about 15-20% of the puzzles from M1e set. Some of the puzzles I do not analyze at all - just "click at the square you have been doing all the time" ;) :).

    Now the question arises: should we just make a correct move or should we be 100% we know (and understand) WHAT is the correct (best) move and after that - make this move? In other words - if I can solve the puzzle without thinking (but not being 100% sure of its correctness) or I can solve it much slower, but being 100% - which move (way of solving) do you recommend? Are any of these better or worse? I am REALLY curious what is your opinion on that matter. Thanks a lot for your hard work, engagement, dedication and sharing all your thoughts! It's great being a part of our tiny chess-ideas-testers community! :)

    ReplyDelete
    Replies
    1. There is a lot to say about memorization. A few quick points first:

      *My gut feeling tells me that memorization isn't good because it isn't a skill.
      *Aox is convinced of that too, and he has gathered some data about it. As always I have questions about that, but that's just a habit of me.

      Yet I doubt if it is a long lasting problem. Since the memories fizzle out over time, while the skill tend to last. So it might temporarily seem to boost your performance, time will correct it. If you are aware of that effect, there is no real problem that needs a specialized fix. Simply do the same exercises once memories have faded away.

      Delete
    2. Should we be 100% sure making the correct move?

      I give a temporary answer: no. We are trying to improve the "Troyis way", i.e raging like a madman without much of a strategy. You will see that when you start slow solving, you soon notice common features in the position, and your solving speed will increase since it makes no sense to keep being slow when the answer already pops to your mind. The gap between the slow tempo and the fast tempo soon decreases, en after a few hours you will be raging like a madman again who totally has forgotten his good intentions of being slow.

      You can only be 100% sure when it is verified by conscious thinking. Conscious thinking takes time, and hampers the execution of skill. It are the disruptions I describe in this post. Switching between thoughts and skill.

      Thinking here leaves nothing behind. When the exercise is over, the thoughts are gone. It is good to remember your goal. I want to get "perfect vision" of the current position. Only that way I can ever hope to see future positions before the minds eye. If it takes me 30 seconds to recognize a pin in the current position, I will never be able to manipulate future positions in my mind.

      So the pins must reveal themselves "immediately". Otherwise the burden on STM will become too big, and I will perish in confusion. So for now, manipulating attackers, attacking squares and defenders in the mind " Troyis style" should be the goal. Later on, a good thought process should be added, making sure that the popping ups are indeed 100% correct.

      That being said, you have to slow down from time to time, and analyse what is going on. Identifying what hinders your progress.

      Delete
    3. The bad thing of memorisation is that you learn the WRONG thing
      You recognise a positiion because of some uncommon things like say: a king a d4 and say at this puzzle we have to click at a1 and h1

      To memorise that is simply wrong, such a information will be misleading.
      I still think that a bad performance is partly based on wrong "knowledge(memorisation)"

      Delete
  3. Good news! I thought our/my goal is to solve as many puzzles as fast as possible - and see (compare) if there is a SIGNIFICANT difference. And now I am solving puzzles quite slow as I want to analyze what is hidden in the position (and what factors I cannot understand or spot correctly). And when I analyze the time factor is no longer at the highest level possible.

    What then should our goal be? If not the speed/pace (tempo) of solving? Understanding? Correctness? Ideas? I am a bit confused.

    BTW. I am not sure, but we have probably missed one trivial subtask! Find the KING! Many times I lose precious time to find the King among a lot of pieces. And I am NOT kidding! I do not claim recognizing the King from the other pieces, but to spot when it is located (immediately). And the puzzles will be really trivial - even kids could do it ;) :).

    Tempo: are you sure it would be MUCH better if I solve 1000 puzzles with the speed of 20 MPM with 95-100% understanding than against 40 MPM with 85% of understanding? Aox - what is your opinion? Can you share it?

    ReplyDelete
  4. @Aox: "You memorise unconciously."

    That happens regardless of the protocol that is used for training. An important question is whether the training produces short-term or long-term benefits in overall chess playing skill (NOT knowledge). A related question is how to measure the long-term improvement (if any) in overall chess playing skill (NOT knowledge) as a result of doing the specific training.

    @Tomasz: "Now the question arises: should we just make a correct move or should we be 100% we know (and understand) WHAT is the correct (best) move and after that - make this move? In other words - if I can solve the puzzle without thinking (but not being 100% sure of its correctness) or I can solve it much slower, but being 100% - which move (way of solving) do you recommend? Are any of these better or worse?"

    Based on my own research into effective long-term skill training methods, I strongly hold the OPINION that understanding what is the correct move (and WHY) is much more effective LONG-TERM than "guessing" without certainty. Guessng is a waste of time for training purposes. This is one of the reasons that I believe "working in the salt mines" for hours at a time trying to get a higher average "score" in terms of X problems per minute is NOT effective for the long term. For learning to be effective long-term, it is important for it to be "effortful," which cannot happen if there is massed practice (the salt mines) that is so rapid that no time for reflection is possible. Without reflection, there is little that gets moved to long-term (skill) memory.

    In this regard, I seem to hold a distinctly minority (of one?) opinion on this blog. That is why I have not commented on the current flurry of activity in salt mining. Even presuming that part-task training is effective (and I think that it can be), massive quantities of practice enables the mind to correlate (short term) and gives the impression of mastery, which is instead familiarity with the specific task(s).

    A relevant question for all salt miners: how do you measure long-term skill improvement after digging so religiously into the salt, and how do you correlate that measured improvement back to the various aspects of training? Or is measurement of long-term improvement considered to not be relevant?

    Please note: while I may be skeptical of the efficiency of salt mining per se for long-term improvement of skill, I have nothing but the highest admiration for those of you willing to invest so much time and energy into investigating various methods of improvement training. There is real value in the research effort, even if it only points out areas that do NOT produce improvements. Temposchlucker and AoxomoxoA have both made significant contributions through their research over the years. As Thomas Edison reportedly said, "I have not failed. I've just found 10,000 ways that won't work." Only several thousand more ways to tried to produce improvement!

    ReplyDelete
    Replies
    1. I would like nothing more than wholeheartedly AGREE with you, Robert, since it is the most logic way to look at things, and logic is my inner idol.

      Yet there is one horrendous little nagging fact that I would like to sweep under the carpet, if Aox would let me, and that is the fact that the only measurable long term effect of 12 years of investigation that could be traced backwards to the origin, is the improvement by doing Troyis.

      I did Troyis like a madman for a few weeks, and years later Phaedrus presented me a knight exercise that was used to measure your rating. Grandmasters could do it fast, Fide masters were less fast, amateurs were horrible. I scored a rating of 2600, like a grandmaster.

      So I'm obliged to settle this question once and for all: can the Troyis-method be applied to other regions of the game? That is for me what all this salt mining is about.

      Troyis gave me the skill to manipulate a knight in my head in a limited space (I'm talking about the board now) at lightning speed. It didn't provide me with a 100% correct strategy to do so, it just provided the skill to manipulate. That's why I am not bothered with correctness, but in the skill to manipulate attackers, attacking squares, targets and defenders in my head.

      Delete
    2. Weeell, the chain of arguments goes like this:
      We want to improve in chess and we dont
      So we try to find trainings methods and key skills for improvement
      We chose tactics, its a key skill, there are many tactics servers to measure our skills in tactics
      But.. we dont improve in tactics, after a initial phase of improvement ( 1000-4000 attempzs) we reach a ceiling where improvment is soooo small that you cant "see" it anymore
      And we all did read tactic books, watched tactic videos, solved tactic puzzles, did learn sets of tactic puzzles and and and

      So I am analysing my ( and others ) non progress in tactics since several years. I did see that a progress in tatcics is possible AT PUZZLES SEEN BEFORE but not in puzzles not seen before, so i was trying to find a logical subset of tactics where i would be able to improve at puzzles i did not see before . This Set was a Set of easy mates in X. But i did improve at puzzles i did see before and not in puzzles which i did not see before

      I did read several scientific papers and found that GM's can do easy tasks at higer speed than IM'S and they are faster than FM's and so on. One of these tasks is "to recognise if a position is Mate or not" . If i Remember that datas correct a GM can tell that in something like 145 ms. Such quick times show that GM's can see chesspositions " as a face" , they did reprogram brain regions which are usually responsible to recognise faces.

      Now we know how GM's get such skills, they do it by delibertate training ( in young age ) , what they are doing is saltmining within their games , within their analysis...
      They look for all checks, look for all escape square OVER and OVER and OVER again.

      What we try to do is to compensate our age and compensate our lack of deliberate practice and apply the medicine concentrated.

      We already KNOW that your idea of improvemet dont work with adults

      We know that : as better the player as faster he can solve a tactical puzzle
      In general we know that we need to get twice as fast to gain 100-200 elopoints
      M1 is just a small subset of all tactics.. if we cant improve in m1.. we dont need to try to improve in tactics.. then we know we cant improve in chess decisiv ( Good player dont make many blunders ).

      You may join the m1-saltmine and try to improve by understanding.. then we may see if that might work too

      But we try to hardwire the board into our brain.. gaining better borad vision and hopefully better tactical vision too

      Delete
    3. The skill how to ride a bike differs from the ability to find the correct road.

      Delete
  5. What is more important if you want to ride your bike home?

    Plese see all of these short videos:
    https://www.youtube.com/watch?v=3PycZtfns_U&list=PLhXK4-QMA8o6JSGRRNFfvEDC7rj3JDwI1

    Such a teaching method is not common in the western world anymore but still: its effective

    ReplyDelete
    Replies
    1. oops the most important sequence at the end is missing: see here : https://youtu.be/O-qesAt92Jw

      Delete
    2. naa.. focus on the most simple subtasks and learn to do them automatic

      Delete
    3. @Aox: "Such a teaching method is not common in the western world anymore but still: its effective"

      Having been trained in Isshinryu Karate-Do by 7th degree black belt Dan Montgomery, I can attest to the effectiveness of the 'Miyagi Style' of instruction. Each sub-skill is isolated and practiced separately, and then integrated into a greater whole, eventually reaching the level of a complex traditional kata (form). The guiding principle is: TRY TO FIND THE SIMPLEST RESPONSE THAT ALWAYS WORKS! Break a technique down into its component parts, work the 'salt mines' on those component parts while simultaneously interleaving integration of those parts as natural responses, and (eventually) folding those constituent parts into more complicated patterns. Skill is NOT the same as knowledge; "know how" is not the same as "know what."

      I started martial arts training at 40 years of age, with no prior martial arts training and mediocre (at best!) physical condition and skills. (At that point, I had been riding a desk in a cubicle farm and programming for over 20 years.) After 17 years of regular (2-3 hours per day, 3-5 days per week) training (and teaching, and getting "teacher training"), I have reached an earned 4th degree black belt level with an awarded honorary title of Renshi (Master of Technique). Perhaps this is the reason that I believe that adult learners CAN improve significantly in just about any field (including chess), under the right conditions of instruction (an expert TEACHER is a vital necessity; most people are unqualified to teach others [or themselves] anything) and dedicated focused practice over a prolonged period of time. Eventually, you reach a level of skill in which you confidently "know" (and know that you can "do") the correct thing in a wide (almost unlimited) variety of situations, without being able to verbalize it in a logical sequence unless considerable effort is expended thinking it through slowly, step by step.

      My views on effective instructional methods are largely informed by my own experience as an adult learner and teacher, supplemented by considerable study outside of academia. Since I view chess as a combative art (albeit one that is primarily cerebral), I see no reason why the same instructional method(s) would not work to produce a skilled chess "warrior," even for adult learners.

      Delete
  6. we need both to get somewhere. I divide the tactical skill in to parts, the quick and the slow part. The quick one is to spot easy tactics emediatly, to perform well on fast and easy tactical problems. The slow one is the deep calculation ability, the ability to still see whats going around at deep calculations, an efficiant thinking process and so on.
    We know good player can do both. A super GM will play better bullet than we play an 5 hour game. I suspect both skills to be related to some extend

    ReplyDelete