Thursday, September 12, 2013

The initiative

Writing in english again is somewhat uneasy. I feel rusty both write- and chesswise. I did read a few old posts of mine. Especially this one caught my attention. I never really put the ideas from that post into practice. But pruning branches and  excluding move orders based on reasoning about the initiative is a great idea. I will experiment with it. All other ideas failed, so there is nothing to loose.

I did a little test with solving an old high rated problemset. It turns out that most of the times I remember the moves, not the diagrams. Since the initiative is much more important than geometric patterns (see the old post I mentioned), it's logical to have a closer look at it.

Further I use a mind mapping tool lately, which is a great way to organize knowledge.

23 comments:

  1. Glad to see you resume posting! There is a wealth of information on your blog. Thank you!

    ReplyDelete
  2. I noticed:
    a) We remember the main variation moves only. All the sub-variations of a puzzle are lost.

    b) the main variation can be quicker calculated when we know it.
    It is easier to follow footsteps. Calculating known patterns is much faster than unknown patterns. We can calculate much better what we expect.

    c) it does not matter how long it took us to find the solution. All hard thinking is in vain. It does not improve our thinking or calculation.
    You summed it neatly up: No DIY! After a minute of thinking we can look up the solution. We dont need to find it ourselves. And that we didnt do a lot of thinking does not matter.
    A lot of thinking has nothing to do with improving our thinking.

    d)Is it all about pattern recognition? Yes, it probably is (if you define pattern not only as s.th. geometrical, but also include guidance, typical rules like "a knight on the rim looks dim". And also include boardvision (=hanging pieces, possible attacks, possible checks).
    However, one trouble is, that similar patterns interfere with each other. Actually there is no interference of patterns, but we simply have a set of dominant patterns which suppress other known patterns.
    We dont apply rules/patterns we learned (because they are rather seldom). We know these patterns, but while solving the puzzle we did not recall the relevant pattern but only recalled the wrong pattern. There is no interference in patterns. It is one pattern (the wrong one) we think of, and the other (the right one) does not interfere - we simply didnt consider it at all.

    e)knowing a pattern and applying a pattern seems to be 2 different things. Even knowing a pattern very well does not necessarily make us using this pattern.
    I found no cure to this problem, but can only suggest what I previously propagated:
    It might help if we know our patterns inside out. Go for easy puzzles, because point a) clearly indicates that we only memorize main variations. Easy 2 and 3 movers hardly have anything else than main variations.

    But nevertheless it is no guarantee that we are going to apply what we learned.


    ReplyDelete
  3. Hi Tempo

    Starting to play and study chess again.You might find this interesting.

    Which speaks the rules of attacking play.

    https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=3&ved=0CEAQFjAC&url=http%3A%2F%2Fwww.qualitychess.co.uk%2Febooks%2Fattackingmanual1.pdf&ei=nwpIUp_UJI-x4APAlIDYBw&usg=AFQjCNH06QRU698xg9OGJapvSupHrOC7KQ&sig2=FajkYQoL0yPO71m0TXApbA&bvm=bv.53217764,d.dmg

    games collection here
    http://www.chessgames.com/perl/chesscollection?cid=1020827

    ReplyDelete
  4. Better version of the game collection found here.

    http://www.chessgames.com/perl/chesscollection?cid=1020635

    ReplyDelete
  5. Hi tempo,

    If you look at my chesstempo Blitz rating graph, you will on the rating number graph a plateau.

    However, I realized that the percentile graph differs.
    We seem to have some deflation at chesstempo, and during the last 3-6 months this deflation was intense.
    In the percentiles I did improve. And "coincidently" I had a sharp rise after I did my training of 11.000 repetitions in "check mate in 2" in the easy range.

    My old ATH from March 2012 was 1936, and it gave me a "better than 89.1%".
    Now I have a rating of 1922, and this gives me a percentile ranking of "better than 91.7%".
    2.6% difference means approximately 50 ranks, or about 30-35 rating points difference.
    Adjusting this means, that my old ATH of 1936 is nowadays worth ~1900, only. And thus it means I improved tactically.

    The interpretation is nevertheless a bit troublesome. Back in March 2012 I suffered much less from duplicate reward reduction. At least I claim I suffer from that, because other users might actually profit from the duplicate reward reduction. It is a "one size fits all". Younger players with better memories need less repetition to remember puzzles, and hence the rating reduction isnt big enough, whereas people like me, you and Aoxomoxoa tend to suffer from this reduction scheme. We would need to have a "first timers" only rating at CT, but then we would still have inflation or deflation issues with the rating.
    The recent deflation might have been caused by people creating new user names, who knew the puzzles already but dont suffer from duplicate reward reduction.
    Or the player pool just became weaker. But somehow I doubt this.

    Anyway: I might not have used the best way to improve, but what is important is:
    It did improve my tactical abilities. And it isnt "my" way, but I got the ideas from you and followed them.
    If you look back what ideas you havent tried - well, for some reason you lost faith in your idea of doing lots of easy puzzles with clear patterns.

    Certainly there is s.th. like deminishing returns. And probably you have seen 99% of all patterns there are to be found.
    However, it is not only knowing about patterns, but it is about: How quick can you apply these patterns? You might stuff your memory full of patterns (and that is certainly important). But what matters is the time it takes you to retrieve these patterns.
    It is much less about active calculation, but far more about the retrieving speed of patterns, which aid "trial and error".
    There might be several patterns shouting at once, while some of those are misguiding you. one might be so dominant that the other patterns wont come to your concious mind. But: the right patterns will eventually be triggered, too. Especially if the retrival speed is fast.
    I noticed this during my own thinking: I found a pattern, tried to make it work, and suddenly I found a different pattern, and knew that I was following the wrong one.
    It happens. But important is, that the right pattern gets it into your conciousness, that it interrupts your thoughts on following the first (and wrong) pattern that came to your mind. It is practice, practice and practice.
    We may understand how to juggle with 3 balls. And we might be able to throw them 4-5 times. But to juggle for 2 minutes, you need to have it "in your blood", and not only to know about juggling.

    I guess you abandoned your ideas about Trois, forks & Co too early.




    ReplyDelete
  6. PART I:

    By way of background: I'm 65 years old. I've been away from the rated chess wars for quite a long time (last tournament in 1975 - USCF rating OTB 1810). I still go through periods of studying chess, for my own amusement and enjoyment, as a Life member of USCF. There is a small local chess club, meeting once per week, with some Class A (and down members). I've gone there a few times, and gotten persuaded to enter their Quick Chess tournaments. In my last tournament (7 minutes + 5 seconds per move, I think), I defeated a FIDE Master - then promptly lost to lower rated players. The losses were pretty directly the result of being in considerable shock at having (finally!) defeated a FIDE Master.

    So, what have I been doing to retain or improve) at playing? A lot of study and thinking about thinking processes. The thinking process study was triggered by Dan Heisman's book The Improving Chess Thinker. However, one of the most helpful insights was gleaned from Emanuel Lasker's Manual of Chess. In the section on Combinations, he describes several motifs - which are NOT the typical tactical motifs, such as forks, pins, skewers, double attacks, double checks, etc. Instead, the "motif" is described in general terms not necessarily using chess terminology. The most important motifs are (1) the encircling motif; (2) the geometrical motif; and (3) the function motif.

    The encircling motif involves two aspects: a target that is immobile and an accumulation of superior force against that immobile target. (Think Dr. Nunn's "Loose Pieces Drop Off!")

    The geometrical motif involves the typical forks, pins, skewers, etc. but with a crucial distinction: the attacker looks through all obstacles (same-side and opponent's pieces) AS IF THE OBSTACLES WERE NOT THERE.

    The function motif is described as being the most important. AS soon as a piece (or Pawn) is given a specific function to perform, it LOSES a certain flexibility, capability, potential.

    Lasker strongly recommends studying and internalizing the various motifs. Ideas (combinations of motifs) are unique to a given position; however, the various motifs are part of technique that should be remembered.

    What does that have to do with chess improvement?

    Crazy (ain't it obvious?!?) Bob

    ReplyDelete
  7. PART II:

    What does that have to do with chess improvement?

    A study of the literature on cognitive psychology as related to chess playing (and there is a LOT of literature on it, especially in conjunction with creating chess programs which attempt to simulate human thought processes) indicates that (uncritically), a Class A player utilizes approximately 500 "patterns", an Expert utilizes approximately 10,000 "patterns", and a Master utilizes approximately 300,000 "patterns" - with NO real definition of what constitutes a "pattern" except the FUNCTIONAL relationships between the pieces and squares in a given position. BTW, the range of patterns supposedly required for Master-level play is estimated at 10,000-100,0000 by Chase and Simon. After considerable research, I found that their estimate was based on the MAPP program, which simulates thinking. In my opinion, that number is as bogus as the statement that "79.3% of all statistical references are made up on the spur of the moment."

    Cognitive research has shown that (regardless of what is considered to be a "chunk") that it takes at least 6-8 seconds of PERCEPTION, followed by up to one minute of ABSTRACTION, in order to establish a "chunk" in long-term memory (LTM). Any time shorter than that will NOT result in storage of a "chunk" in LTM.

    There are some bad implications for these findings. For example, the MDLM "7 Circles of Hell" are virtually guaranteed to NOT work, simply because the LTM is not being filled with relevant "chunks" (even if you were studying EXPLICITLY the relevant "chunks" - which you are NOT). The same reasoning goes for tactics servers such as Chess Tactics Server of Chess Tempo.

    As human beings, we gain considerable power from the ability to abstract from the specific situation to a general "rule" (which may or may not be applicable to a new specific situation). Without intensity of study and total absorption (and time to absorb), it is highly UNLIKELY that permanent progress will occur.

    Consistently throughout most high level chess literature, there is the admonition to study complete games contested by the highest rated players. A deep study of every move within a given game will add more "chunks" to LTM than 1,000 simple tactics puzzles done without thought.

    I've used Chess Tactics Server to maintain my tactical "eyes". However, I pay no attention to the time. If I can "see" ALL of a solution when I first see the puzzle, then I will click out the solution. If I don't see the solution immediately, I ignore the time (and thus lose rating points regularly), and study the problem until I understand everything that exists in that position.

    A couple additional sources for emphasis on the importance of "functional" thinking while playing chess are Botvinnik's book Computers, Chess and Long-Range Planning and Lenhares' program Capyblanca.

    Hopefully, that will provoke some more thought and study. In any event, thanks for reading!

    Crazy (ain't it obvious?!?) Bob

    ReplyDelete
  8. MDLM 7 circles of hell doesnt work because of other reasons, but... never mind.

    Anyway - this does not mean, that learning from puzzles is impossible. You write "same reasoning goes for....chesstempo". That is a strange conclusion to make. It depends how you use it, doesnt it?

    I think that the key to successful learning is to break down the complexity of chess. We need to simplify things to learn them: "encircling", and "geometric" and "function" is such an attempt to break down complexity. It is a good attempt.
    However, breaking it down to even more is possible, too.
    At chesstempo you can filter for "fork", "pin", "discovery", etc., and you could also filter for "mate in 2", or for tactics that happen in the first 10 opening moves in a specific opening.
    All this simplifys chess more, enabeling us to learn each step at a time.


    I studied a book call "practical rook endgames".
    It explains typical tactical motifs and gives examples in games of these motifs. Even though there is almost only one example for each typical tactic, I absorbed them very well and was able to implement them in my own games.
    I reread this book over and over several times so far, and am still not finished with it.
    There is certainly some truth in your argument for perception and absorbtion. And it makes me think of "quality" (my rook endgame book showed just a dozend of motifs) is more important than "quantity" (tons of puzzles).
    However, both seem to work, as long as you keep repeating what you learned. Otherwise we forget.

    The most troublesome part is actually not knowing patterns (=learning and getting them into your LTM), but the hard part is the retrieval the gained/acquired knowledge and apply it in your own games.
    I like to compare it with learning languages: you can learn thousands of words and have a rich knowledge of vocabulary. However, actively speaking the foreign language is very difficult.
    In a foreign language conversation, you might end up of thinking to long, searching for the right words and building the sentense you want to say. In the meantime, conversation went on, and your sentence arrives in your brain far to late. Result: you are very silent in such a conversation.

    Instead of learning a lot of vocabulary, it is useful to have learned full sentences, which you prepared at home before a conversation starts.
    Transfered to chess, I agree with you that we learn a lot by memorizing not only the tactics, but how they emerged in a game (full sentences). So memorizing full games could be helpful in actively placing your learnt tactics in your own games.
    However, there is truth in both: Learning full games is useful (="full sentences"), but of course it is important to know "vocabulary", too. Without vocabulary, no sentences.
    For instance, when I learned english I had this sentence memorized:
    "If it had rained, I would have taken my umbrella". Now I use the grammar but replace it with vocabulary, and hence the sentence I use in a conversation is for instance "If I had missed the bus, I would have stayed at home".
    So both is neccessary: lots of tactics (repeat them! You repeat vocabulary, too, when you learn a language), and trying to apply them in a context (learning full games, or at least how the tactic emerged 2-3 moves before the tactic starts).
    I recommend tagging puzzles after you solved them. A feature you have at chesstempo. It seems to help to organise patterns in my brain, making it easier to actively recall them later.

    ReplyDelete
  9. I think we are more in agreement than in disagreement. I certainly agree that there is something of value which can be obtained from tactical practice on sites such as Chess Tactics Server and Chess Tempo. The point (to ME) is to gain as much long-term benefit as possible, rather than short-term benefit, in the most efficient way time-wise. I do qualify for the "old dogs" category of ADULT chess improvers!

    If the goal is to do better at Chess Tempo Blitz, then memorization of as many patterns as possible may help improve that rating. However, I don't think that it will significantly improve your OTB skills in tournament conditions.

    IMHO, the conceptual notions of "encircling", "geometrical", and "function" are not an attempt at simplification, but are a form of conceptualization, i.e., generalization from experience. If one does not analyze followed by synthesize, then no useable long-term memories are established that are plastic, i.e., that can be used in different situations.

    Charness in his article "Human Chess Skill" (in Chess Skill in Man and Machine, Frey) states that it might be possible for an unskilled chess player to memorize 10,000 tactical positions. The result would be that you could perform the typical recall process (i.e., when shown a typical chess position for 5 seconds, then copy that position to another board) as good as a Master - but that it would hardly at all change your actual skill at playing chess.

    In cursorily reading over some of your diary on Chess Tempo, I noticed two things of interest.

    (1) In doing a large number of tactics problems, you experienced motivational burnout. At that point, I suggest that whatever value might have accrued from exposure to so many different positions disappeared. The large number of positions (16,500) in approximately 48 hours allows for no more than approximately 12 seconds per position. That is insufficient time (according to cognitive research) to store significant "chunks" (as related to chess skill) in long-term memory. There must be sufficient time for perception (at least 6-12 seconds), followed by 1-? minutes of abstraction/generalization in order to establish a meaningful "chunk" in long-term memory.

    (2) When reversing the board position (upside down?), it became evident that you had not gained significantly using the same puzzle set. The positions "looked" different, and consequently did not trigger an association with the patterns you worked so hard to acquire.

    In my own (much more limited) experience, I have gained considerably more actual playing skill from knowledge coupled closely to understanding and then generalization of concepts from that knowledge. Memorization is great for regurgitating rote information on command (such as on tests in the academic environment). Memorization is NOT as useful when faced with new situations which require a conceptual framework in which to apply knowledge and experience.

    Lasker put it rather succinctly:

    Of my 57 years, I've applied at least 30 to forgetting most of what I've learned or read, and since I succeeded in this I have acquired a certain ease and cheer which I should never again like to be without.

    It doesn't seem to me that Lasker put much stock in rote memorization.

    All the best.
    Crazy Bob

    ReplyDelete
  10. FWIW (not much): On Chess Tactics Server, I've "solved" 25263 problems with 80.0% success. My current rating is 1550, with my ATH at 1634 (5 years ago). I actually find higher rated problems easier to solve than lower rated ones.

    I read MDLM's book on Rapid Chess Improvement but (so far) have resisted any inclination to attempt to mirror his experience. I'm might be "Crazy" but I'm not INSANE!

    Crazy Bob

    ReplyDelete
  11. There is a correlation that a high CT Blitz rating corresponds with a high Fide elo. The main trouble isn't to get a transfer from an increase in CT Blitz to real OTB. The real trouble is to improve in CT Blitz!
    What I am after is actually not "just" improving in CT Blitz rating, but to improve on puzzles I see for the first time ("first timers" or "non-duplicates"). Chesstempo fortunately has the filters to check for that.

    For CTS I have hardly any experience with it, but it is not comparable with chesstempo.
    CTS is a waste of time. You have no options to filter for certain puzzles, and you cant repeat puzzles in own tailored puzzle sets. I guess, with you it is the opposite: you hardly have experience in the many filter options you have at chesstempo (which you actually only have if you are a paying chesstempo member).
    With 25K puzzles, you wasted probably a lot of time, since you hardly repeated most of the puzzles within a few days. The gap between having seen a puzzle and having seen it again was probably too big most of the time. If you dont repeat what you learn, probably it is not even worth starting to learn it, because you will forget it later anyway.

    For the perception time, I dont really know what it means. However, I'd dare to say that of course you can quickly tell if the traffic light switches from green to yellow to red. That does not take 8-9 seconds.
    Also learning s.th. does not necessarily need a minute. As a child I touched once an electrical loaded cow fence. Boy, I NEVER do that again!
    I learned s.th. in almost no time, and still know it.
    So whatever theory that is you are referring too - it certainly is written in some specific context.

    As for chess puzzles, it happens that I had solving times for 2-mover puzzles within 2 seconds. That is perception time including mouse movement and clicks. I mean, how can this be impossible, if I evidently can do this?

    My last training was trying to improve retrieval speed from known patterns ("check mate in 2" puzzles).
    It is a bit funny, that you tell me in point (1) that it is impossible to learn them (according to research) - while I learned them!
    I guess you need to read that source again where it talks about perception time and absorption time.
    Of course the longer we look at s.th., the better it sticks in our mind. But I kindly remind you of the cow fence I touched, and the traffic light you need to become aware of. You have little time for perception in such cases - and still it works.

    Point (2) is indeed facinating:
    I am not sure, why I have trouble to solve puzzles up side down.
    That is not only valid for puzzles I memorized, but also for "first time seen" puzzles.
    My solving time gets down, no matter if I have seen the puzzle before or never have seen it before.

    I was trying to find a comparisson.
    I might have found one:
    Try to read this sentence backwards (from right to left, like you would read it in a mirror):

    "Was it a cat I saw?"

    So after you managed to read it backwards, you might smile at what the sentence means backwards.
    The point is, that reading it backwards takes more time to do, despite it is somehow still the same.

    About MDLM - He might not have improved. And I believe his method (I only read indirectly about it) cant work. >>> I <<< believe he cheated, and his book is only stipulating its readers.
    Because you mentioned him here again: I dont think tempo, or aoxomoxoa or bright knight or james stripes take MDLM seriously, and of course we dont follow his methods.

    ReplyDelete
  12. Point (1) was in the context of absorbing "chunks" (functional relationships between pieces and squares in a given chess position) into LTM, not in stimulus/response scenarios such as peeing on an electrical fence. That you can remember NOT to do THAT again is because of the PAIN that is burned into your lizard brain, not your cerebral cortex. That is NOT and does not require a high level cognitive recognition state or a long-term memory based on "chunks".

    Point (2) would seem to indicate that your memory of the "patterns" you have seen is more of a visual-spatial memory rather than an abstraction of pertinent functions involved in chunking. The theory of "chunking" shows that the pertinent "chunks" (I am deliberately avoiding the term "patterns") are essentially abstractions, without specific visual-spatial information associated with them. There is a considerable literature available online, and easily found using Google.

    It is most unfortunate that the terminology of "patterns" and "chunks" tends to blur into each other. "Chunks" (as used in the literature) is NOT just specific tactical patterns such as forks, pins, skewers, etc.

    Your goals are your goals; I have no dispute with however you choose to pursue them, and wish you every success.

    Regarding the MDLM contribution to adult chess improvers: if I am not mistaken, our blog host as well as several others initially began as Knights Errant, a direct attempt to duplicate the fantastic ratings increase supposedly experienced by MDLM (and documented in his articles and book). I am also familiar with the allegations of "cheating" against MDLM. I am unconcerned with whether MDLM "cheated" into a higher rating class (and most definitely have no interest in duplicating his results if that was the case). According to most of the Knights Errant blogs, no adult was able to duplicate the extremely rapid chess improvement experienced by MDLM. There may well have been other variables at work, which would have allowed such rapid improvement. If so, duplicating his method should have confirmed it. I repeat: to date, no one has been able to duplicate it.

    The "Deliberate Practice" approach postulates that it takes approximately 10,00 hours of focused, deliberate practice aimed at elimination of areas of weaknesses in order to reach the level of mastery, regardless of the subject. It also postulates that there are no short-cuts available to reach mastery. It is NOT merely a matter of putting in 10,000 hours of "practice." You have to "pay your dues" and that means a lot of very hard work.

    My interest (and the reason for attempting to tease Tempschlucker out of his [hopefully temporary] blogging hiatus with my comments) is in methods that work for long-term OTB improvement for ADULTS within a framework of deliberate practice. This blog has an enormous wealth of ideas that have been tried and (generally) found wanting in terms of persistent long-term chess improvement.

    Thank you for your responses, which provide several points to ponder.

    All the best,
    Crazy Bob

    ReplyDelete
  13. "pattern" is for me everything that can be described and learned.
    It is probably useless to make distinction between different kinds of patterns (guidance rules, geometrical set ups, or even chess rules like "bishops move diagonally"), and usually is only important when things get discussed in great detail in theory.

    I followed the idea of "deliberate pratice" and my interpretation of deliberate practice was: "to concentrated on my fails (blunders) or puzzles I solved too slow."
    However, surprisingly this approach didnt deliver strong results for me. Same was said by aoxomoxoa when he analysed his own data.
    The more successful approach for me was: trying to become faster in what I know. That means repeating everything, not only my failed puzzles but also my puzzles I solved correctly in good time.
    I still believe both approaches ("concentrating on blunders" and "concentrating on speed") are "working" (=seem to pay out in real fide elo), with the surprising tendency that speed matters a bit more. (Empirical Rabbit was right in the first place).
    Trying to improve in speed IS actually deliberate practice, too.

    For point (2):
    I cant rule out, that I have trouble in identifying pattern when I am looking at them upside down. Since I see pattern as everything that can be learned as "rule" or "geometrical set up", "patterns" are probably the main cause for being so slow. But what kind of patterns are troublesome to me?
    First, I tend to still look from bottom to top, trying for some seconds to find a solution for the wrong side. So if it is "black to move" I tend to look for white, and the other way round.
    I also tend to search in the upper half of the board to look for checks. And I seem to have trouble to find the opponent king. It feels to me like trying to find backward moves.
    I tend to need to remind me in which direction the pawns are moving.
    (That is a pattern, too: pawns move forward, from bottom to top!)
    I often need several seconds before I can start checking variations from top to bottom.
    It is like writing with the left hand (for a right handed man), or like reading a book while holding it against a mirror.

    I would not say that it proves that a learning method is wrong or right, but it rather gives insight how my memory works.

    I think it is worth to solve puzzles upside down. I hope it sharpens my "eye" in what is threatened against me. Maybe (?) all these troubles are pretty common and that is the main reason why attacking is always easier than to defend - for a lot of people?


    For the stated 10.000 hours: Wasnt there a study (we probably read the same research paper), that some Masters needed 3.000 hours, whereas others needed 20.000 hours? The average might be 10.000 hours, but the range is pretty wide.
    Either it is inert talent, or there are ways that are better for improvement purposes, and ways that are worse. And that would mean, that there are short cuts.
    One of them (me thinks) is having a coach. However, Magnus Carlsen had no GM as parents, and only later (when he was already close to expert level) he had a coach. It cant have taken him too many hours to achieve master level, because he was to young to be able to do 10.000 hours when he achieved Mastery.

    Deliberate practice feels a bit dry, since it always means working on s.th. that is outside the comfort zone. The approach I use is pretty dry. Thats why tempo said once: "back to the saltmines!".

    ReplyDelete
  14. The FIDE Master title is associated with an Elo rating of 2300+. Magnus Carlsen learned the rules of chess at approximately age 5 (Wikipedia). He crossed the FIDE Master rating threshold at approximately age 12-13. Not exactly 10 years and 10,000 hours of deliberate practice, but pretty close. Using 7 years as a baseline, 10,000 hours can be achieved by investing 1429 hours per year in study, or approximately 5.5 hours per day, 5 days per week (with time off for good behavior ;-) ). For a focused young person, with no other responsibilities (working, family, etc.), that does not seem very far-fetched.

    The literature has a wide range of estimates for number of required hours (2,500 - 20,000) as well as an extremely wide range for number of "patterns" (Class A as low as 500; GrandMaster as high as 300,000; average 50,000 on a range of 10,000 - 100,000). The number of hours of deliberate practice required seems more robust than the number of "patterns", because the hours of deliberate practice numbers is based on linear studies of many different "experts" in a wide variety of fields, not just chess. On the other hand, the number of "patterns" required to perform like a "Master" (at the extremely limited skill of seeing a position for a few seconds, followed by recreation of that position on a different board) is actually an extrapolation (let's be charitable and call it a SWAG [scientific wild-ass guess]) form studies based on a computer simulation (MAPP) and extrapolated from it.

    The crucial difference IMHO is NOT the 10,000 hours figure but deliberate practice. The problem we all face is that NO ONE knows (1) what are the relevant "patterns" and (2) how to infuse those into long-term memory. I would hypothesize that what worked for one individual (Carlsen, for example, or MDLM, as a notorious example) might NOT work for anyone else, or at the very best, might work but with considerable variation in results. Given the billions of neurons and the associated connectivity in and unique to a specific brain, it would boggle my mind if the same pedagogical programme would produce the same results in every student.

    I agree completely that there are some pedagogical methods that are more efficient than others. I believe that is the main thrust of this blog and others like it. The central question is NOT how long it takes a particular individual to achieve mastery, but how to achieve it - period.

    I suspect that there is no magic generalized formula. What works for you may not work for me, and (obviously) vice versa. That stipulated, I think insight into effective training methods can be gained from the available literature on both chess and cognitive psychology.

    Regards,
    Crazy Bob

    ReplyDelete
  15. (1) For the relevant patterns, there is meanwhile sufficiently known.

    I (for instance) can tell you the patterns you need:
    First you need to know the rules of chess.
    Leaving the very basic patterns aside, there are patterns for positional play, and tactical patterns.
    Leaving again positional relevant pattern aside, one might want to know the most common tactics that win material gain of at least 1.75 pawn units. Tactics that happen often in the game.

    "Happen" means, that the winner found the pattern. However, the loser was not aware of the tactic (otherwise he would not have blundered).
    Agreed so far?
    So I can now come to the point:
    "what kind of material winning patterns are most important to know?"

    Or differently asked:
    "Which are the statistically most relevant patterns?"

    Well, there is an answer to that:
    At Chesstempo, you can find these patterns in the range between 1200-1650 CT Blitz rating. (~50% of all chesstempo puzzles are within this range, and all puzzles stem from real games from a 2.5 mio games database, the average player in these games has 2200 fide elo).

    It is a gauss-bell curve, which has its peak at CT Blitz rating of ~1450.
    The far most important theme people fall victim to seems to be "check mate" with a whopping 42% share of all winning tactics, and amongst check mates, especially the "check mate in 2" tactics are dominating (8000 of 45000 puzzles are "check mate in 2"). If we set aside check mate patterns and look at material gains only, then the most common themes are "forks", with about 8000 non-mate-forks (8000/45000 = ~18%).
    Other important non-mate tactics are pins, discoveries, distractions/overloadings, capturing defenders, hanging pieces, but I am currently a bit to lazy to look their share up. Forks are by far dominating though.

    Anyway, it is no coincidence why I trained especially "forks" and "check mate in 2".
    But knowing only the most common patterns will probably not make you a master player. I guess that masters know most patterns. Which pattern one does not know is different for each individual, of course.

    But you can easily find the holes in your knowledge: The puzzles you failed or solved slowly despite the fact that they are pretty easy for others. Yes, they are "easy", but not easy for you.

    I also want to point out, that I believe that training to speed up your solving time for patterns that are known to you, is also a way to improve. For me, this way was even more successful, but I still believe that both (training to become faster AND training what you dont know well) go hand in hand and can be trained at chesstempo with the help of the available SR algorythm (SR = spaced repetition sets. You need to be chesstempo gold member for that, though).

    About "available literature" for chess improvement:
    Just because somebody has written s.th. down, it does not mean it is correct. Example: MDLM wrote a book about chess improvement, but I believe what improved him most was using a chess computer during his games under his long black coat, while most other players were walking around in short shirts because it was so hot. That is was what MDLM forgot to tell.
    Of research papers, most helpful were those of "Campitelli and Gobet". Others were not specifically on chess and requires you to do a lot of interpretation. The sources for research paper can be found at aoxomoxoa wondering blog.

    (2) How to infuse a pattern into your LTM:
    Getting a pattern into your memory is one side of the problem. More troublesome is, however, to retrieve it from there when you need it.

    ReplyDelete
  16. Hi tempo,

    either I have currently a rather good run (we know, that happens, and can last for some weeks, or even 2 months), or (and this I hope very much):

    My "check mate in 2" was really beneficial.
    I reached a new ATH in chess.com rating, I reached a new ATH with "better than 93%" ranking at CT Blitz rating, I reached a new ATH at "guess the move" at CT.

    All of it looks like a "before check mate in 2 training" and "after check mate in 2 training".
    The difference is a sharp rise in strength.

    I think that the "check mate in 2" didnt only improve my ability to look for mates, but (I think) it also improved my vision for important squares. The later is actually what is probably more important.
    Check mate patterns (mate threats) do play a rule in my games, too.

    As always: I am the only testing person. In case your argument will be again "deminishing returns":
    I was able to solve most "check mate in 2" in the beginning, too. The difference is the time. I can find those check mates now a lot faster than before.
    So knowing a pattern, and improving the retrieval speed of a pattern are 2 different things.
    I guess there will be a diminishing return when you are able to solve them within 5 seconds.
    But - are you already at this point for your "known" patterns?

    ReplyDelete
  17. Hi Tempo,
    me again - Munich.
    I improved. Also "again".

    I hope I discovered s.th. important for chess improvement.
    Everything we talked about - it more or less is important. And we did it.
    You did a lot. I did a lot. Aoxomoxoa did a lot.

    But the results seem to be different: I won this season 6 games out of 5, against an opposition of 1966 Fide elo. No loss, just 2 draws (in case you wonder where I lost the one point).
    Let aside how much I improved. Maybe it is just a good run, and statistics will pull me back. But one thing I believe is for certain. I improved. And continue to improve.
    But why? Aoxomoxoa did the same. you did partially the same, and looking back, I think it could work, too.
    So why me?
    And I guess I found the difference. The missing thing. S.th. you probably felt is there but never found it.

    I apply what I learned in my games. I pull the learned material out of my memory and apply it in my games. Of course, if there is nothing to draw from your memory, you cant "pull" (recall/retrieve) it.
    But this is certainly not the case with you. You did more than 100K puzzles, and s.th. must be there.

    The real problem is this: you cant retrieve it.
    It is like learning english in school: it is in the vacation that you start to use some of your passive english knowledge and make it active.
    It becomes active because you are forced to speak actively. But in chess nobody forces you to actively use your knowledge.

    Remember when you had trouble understanding english despite having it learned in school for many years? Chances are, you couldnt follow english movies or have deep conversations.
    But meanwhile I believe you can. (You can, dont you?)

    You couldnt follow conversation or movies, because (yes, sorry) you were too slow. That is pointing into the High speed easy puzzle training.
    But that is not the thing I like to repeat over and over again.
    That is not the missing detail.

    The missing thing that I did, while others did not:
    I wanted to prove I was right.
    I told you that I had "recalls" during my games. I recalled puzzles in my games I had seen at chesstempo.
    Because of this, the desire to be "right" I started to look out in my games if I can recall any tactics I have seen previously in CT. I also experienced it the other way round: I saw puzzles and remembered my Blitz games and thought "oh, I had a similar tactic in one of my games recently". I also tried to recall easy patterns when you presented difficult puzzles. Just to prove everything is based on easy patterns.
    So this "I am right - you are wrong - I will prove it - I will prove it", well, sorry for that.
    But this was what made me use actively my learned easy patterns in my games.
    In that way, I activated a lot of puzzles and made them part of my everyday chessplay.
    Like speaking english.
    I am sure I know much more chess patterns than I actively use in my own games. That is with english the same. My everyday active english vocabulary is just a fraction of what I know passively.
    But I retrieved enough from my chess memory to be a better player.
    You need to look out in your games of what you learned.
    At the start it might be a bit easier to do so if you look for simple tactics, as it is easier for you to use simple sentences in english, too: "My name is Peter - what's your name?". But enough of easy vs hard. Lets leave this aside, and you may rather think with me about how to activate and apply what was previously learned.
    Or lets discuss why it is impossible to activate what we have learned. Or if you think the activation (retrieval/recall from memory) isnt the problem.

    ReplyDelete
  18. I am, but I'm a bit busy with other things. Yet I will react within a few days.

    ReplyDelete
  19. Life is catching up with me, so I'm afraid I will not post again anytime soon. See you later.

    ReplyDelete
  20. Tempo, my going to be ECF grading will be ~188 = ~2150 Fide elo.
    Previously (Last year's season) I achieved 2060.

    To have an increase from 2060 to 2150 was only possible, because this year´s season I managed a performance of ECF 200 (~2250 Fide elo). My former 2060 was an initial start, so my performance was 2060.
    Comparing both performances, I performed about 190 Fide elo better than previously (2250-2060 = 190).

    If you like, I give you per private email my full name and a link to the list where I am listed (I dont like to post such details in the Internet), so you can convince yourself and also to track my going to come improvements in the future.

    I said it before (see above): I believe the one important trick of improvement is: activate your knowledge (your memorized puzzles).
    I could imagine, that a well assorted set of puzzles might help more than thousands of puzzles.
    I am not sure though. I believe I know which puzzles helped me the most for my task to improve in chess. But that might be different for each player.

    Reminds me of the following statistics:
    the first most frequent 25 words make up about one-third of all printed material in English, and the first most frequent 100 words make up about one-half of all written material.
    That means if you know these 100 words very well, you are able to read faster. That is (the/one) trick of speed readers.

    So the need to train thousands of puzzles is maybe not required that much.
    Dont know if that fits here, but do you know the following old saying? --> "A foot-kick in the a... (bottom) tells more than a 1000 words!"

    Here a link to the statistics of "Most_common_words_in_English":

    http://en.wikipedia.org/wiki/Most_common_words_in_English


    If you send me a recent game of yours per board-mail at chesstempo, I can have a look at it and try to find out how you can improve your play. I am not going to tell you where you blundered (any chess engine can do that), but I try to identify where you had probably knowledge which you did not apply in your game.
    I did this coaching already for others, and for some people I was able to identify such things, but admitedly there were also people were I could not find much to comment on.
    I am sure you have stored lots of patterns and guidance rules in your long term memory, so I am full of hope to find a lot to activate for you.
    Tactics (blunders) dont just happen, and there are methods and thinking habbits that promise to be more successful.
    You tried almost everything, so why not trying this with me, too?

    I believe that your motivation is so low, because you tried everything that you could think of, but all of it resulted in very litte improvement.
    One student of mine is aged around 50 years. He did improve a lot (but started from a lower level, too). Not only by solving puzzles, but part of his improvement was probably discussing his games with me, too. He might become a little bit stronger tactically, but his method of playing improved a lot. I can see that from comparing his first games with the quality of play he displays now.

    ReplyDelete
  21. Hello tempo,

    even though I improved tactically, my tactic improvement is not as big as my real OTB improvement.

    Chess is not 99% tactics.
    Sure, tactics is important, but statistical relevance shows that most puzzles have a CT Blitz rating in the range 1200-1700.

    I can see that FMs and IMs and GMs have higher tactics-ratings than me. There is certainly a good correlation of tactics skill and real Fide elo rating.

    So yes, tactics matter, but...

    I improved mainly by changing my thinking process.
    These are guidelines. You found them and gave them a name.
    There are more important guidelines and less important guidelines. I believe one of the most important guidance rules you should constantly tell you during your games is: "to take is a mistake".

    Because if you take an attacking piece, then your opponent recaptures (his attacking pieces is usually guarded) und thus he simply replaces the attacker. But you, as defender, have lost a piece which is not replaced but simply gone.

    Taking loses time. If you get taken instead, your pieces tend to get drawn into the center.

    I used to tell myself during my games "to take is a mistake". I sat on my hands and fought the urgent feeling to take something. Instead I kept the tension on the board.

    I have meanwhile some students, and one 50 year old guy made approximately a 200 Fide elo jump.
    He hardly improved in tactics (though he trains the saltmines), but he told me that he believes that analyzing his games with him helped him most, and it totally changed the way he was thinking and playing. It made him consider much less moves (a lot of takes he considered, but knew that if it does not yield anything, a take is a mistake, so if there is "nothing to gain" he spends now much less time on such a move.

    In case you ever come back to chess training, post me a board email at chesstempo, and we can analyze your games together.
    After all, it was you who helped me to find an effective chess training.

    ReplyDelete
  22. Occasionally I check here to see if you came back. But meanwhile there is about 1 year of silence. Maybe better. Isnt chess a waste of time after all? And about improving in chess: no matter how strong you are, there will be someone who is even stronger. Even if we reach an IM title, there are so many GMs who still are much superior to the average IM.

    But in case you started OTB chess this season again, drop a note via boardmail at chesstempo.

    In any case, this is nice blog. It was the first blog I started to read regularly and it brought me onto the right track. Since new entries dont seem to happen, I started re-reading some posts of yours from the past. We need to repeat what we learnt - otherwise we forget.

    ReplyDelete