Monday, April 29, 2019

How does system I work?

I postulate the following: only system I can make you perform better at tactics.

Rehab of trial and error
That is why I want to rehabilitate trial and error. Since that is the way system I works best. If you need to guide system I by system II, you add slowness to the equation. Since system II is notoriously slow.

When trial and error fails
trial and error fails in practice, for different reasons:
  • repeating the same moves and lines over and over again
  • entering a tunnel and continuing ad infinitum

The actions of systems II that can be safely incorporated
The actions of system II involve the following:
  • Stopping system I when it is repeating itself
  • Stopping system I when it is working in a tunnel with no end
 These actions can be safely incorporated into your solution process, since they don't  require time. Thinking should not be incorporated, since thinking requires time.

Attention
The two actions of system II that don't require time, do require attention.

Where system II is, there is attention. Where attention is, there is the power to discriminate. Where attention is, there is intelligence.

System I is semi intelligent
System I is only semi intelligent. It emulates intelligence. It brings you safely to your office. Even when you didn't intend to go there. Without the attention of system II, it can do silly things in an intelligent looking way. System I can work on autopilot. But it can't learn on autopilot. It continues to make the same mistakes over and over again. We even have a name for this type of "training". We call it the salt mines of Krakau.

It is relative easy to learn to see the material balance in a glance.
It is relative easy to learn to see whether a position is about mate, promotion or gaining wood in a glance.
It is relative easy to learn to see the points of pressure in a glance.
It is relative easy to learn to see the lines of attack in a glance.
It is relative easy to learn to see the function of the pieces in a glance.

But it is not easy at all to remember to look at all five issues above without thinking. And with thinking you add time consumption, mental resource consumption and energy consumption to the equation.


29 comments:

  1. Tempo said:
    It is relative easy to learn to see the material balance in a glance.
    It is relative easy to learn to see whether a position is about mate, promotion or gaining wood in a glance.
    It is relative easy to learn to see the points of pressure in a glance.
    It is relative easy to learn to see the lines of attack in a glance.
    It is relative easy to learn to see the function of the pieces in a glance.


    All these task are done by a better player faster, i guess because of more and bigger chunks? It is NOT easy to reach the performance of a Master at ANY of these "easy" tasks.

    ReplyDelete
    Replies
    1. I don't worry about the differences in speed between system I between persons. I worry about where I am obliged to apply system II. And that is the thought process which connects these 5 glances. Since that thought process uses system II. And since system II is so slow, it eats resources. 95% of the time, for instance. If I can learn system I to do that job that is now done by system II, then I hit the jackpot.

      Delete
  2. PART I:

    Temposchlucker states:

    The actions of systems II that can be safely incorporated

    The actions of system II involve the following:

    Stopping system I when it is repeating itself
    Stopping system I when it is working in a tunnel with no end

    . . .

    Attention

    The two actions of system II that don't require time, do require attention.

    Where system II is, there is attention. Where attention is, there is the power to discriminate. Where attention is, there is intelligence.


    While revisiting Dr. Daniel Kahneman's excellent book Thinking: Fast and Slow, I found this:

    Other mental activities become fast and automatic through prolonged practice. System 1 has learned associations between ideas ...; it has also learned skills ... Some skills, such as finding strong chess moves, are only acquired by specialized experts. ... The knowledge is stored in memory and accessed without intention and without effort.

    [AND, it's FAST!]
    ...

    ... attention can be moved away from an unwanted focus primarily by focusing attention on another target.

    In the martial arts, there is a concept of "I did not do anything: 'it' [System 1] did it." Through countless repetitions of the same movements, System 1 acquires the capability to respond to subtle "cues" without conscious thought and to provide an appropriate response. System 2 is not notified in advance and is unaware of "danger" in the situation until AFTER the response has already been made. There may be a feeling of "Surprise!" after the situation is resolved. Usually, however, there is just a feeling of being "disembodied" during the altercation, observing the actions/reactions with full attention while the body performs the ingrained response. System 2 does not interfere nor attempt to direct the step-by-step actions taken.

    There is a description of how Vassily Smyslov played chess. Different people have commented about the accuracy of "Smyslov's hand": he just moved his hand and his hand "knew" which piece to move and where to move it.

    This is the ideal state to strive for while playing chess. It is based on the intuitions of System 1, N-O-T the logical thought processes controlled by System 2.

    Quoting from Dr. Kahneman's book again:

    The psychology of accurate intuition involves no magic. Perhaps the best short statement of it is by the great Herbert Simon, who studied chess masters and showed that after thousands of hours of practice they come to "SEE" the pieces on the board differently from the rest of us. You can feel Simon's impatience with the mythologizing of expert intuition when he writes:

    "The situation has provided a CUE; this 'cue' has given the expert access to information stored in memory, and the information provides the answer. Intuition is nothing more and nothing less than recognition."

    ReplyDelete
  3. PART II:

    Part of the problem of adult learning is a learned reliance on logical, step-by-step processing for solving most issues (including how to train System 1 to play better chess). This is one explanation for the strong suggestion (from various chess trainers and masters) of following a “thought process.” However, the proponents focus on elaborating the REQUIRED individual steps without stressing the importance of hammering it into memory until it no longer requires conscious thought (System 2) but is automatically followed without being aware that you are even following it. If we can regurgitate the steps of the process on command from long-term memory, we ASSUME that it has been incorporated into System 1. That is just not true! Just as in martial arts, it requires “countless repetitions of the exact same movements” IN RESPONSE TO SPECIFIC CUES in order to build the requisite long-term memories.

    So, not only is an ingrained “response” (thought process) required, but there must be an association with the “cues” that trigger specific (and appropriate) responses. This association is not formed by “solving” countless tactical problems but by building awareness of the relationship between the “cues” and the appropriate response. There is no conscious thought involved, such as “When I “see” THIS, I play THAT.”

    However, to reach that blissful state of “it” playing the moves, countless repetitions of the association process must occur.

    I think I’ve had some success in this regard. I previously noted my training process of drawing lines of relationships (LoA and PoP ) on tactical chess diagrams before attempting to solve them. I’ve done this on several books over many months. I don’t try to solve the problem until after I have looked at the various relationships in this way. I don’t make any attempt to memorize specific problems or solutions. I just work specifically on “seeing” what is on the entire board FIRST. Usually (not always) that is sufficient to provide the needed “cue” to solve the problem. If I don’t get the problem right, then I try to work out what I overlooked as a “cue”.

    As a consequence, I have measurably improved my tactical skills. For a long time, I could not advance beyond 1500 on any chess tactics server: CTS, Chess Tempo, lichess. CTS no longer exists, unfortunately. Currently I am somewhere between 2000-2100 on lichess (NOT blitz mode) and between 1750 and 1800 on Chess Tempo (blitz mode); both of those rating have steadily improved over time. My USCF Quick chess rating was below 1500; it is now 1642 and rising; I won my last tournament game last night, which has not been included in the rating yet. I have noticed a considerable increase in awareness of tactical shots during my games. Previously, I primarily relied on strategic ideas and stock endgames for grinding out wins because I just couldn’t “see” the tactical ramifications. I now play much more dynamically and concretely, and have uncorked some great tactical shots to finish the game or to save a game. I’ve played several games in which I played opening variations that I have neither seen nor studied. Analysis of the games by GM Stockfish shows that I’m “seeing” the major tactical ideas that lead to specific variations.

    I believe it is NOT the thousands of problems solved that is at the root of the improvement. Instead, I think it is the ingrained process and the subconscious storing of the associations between the “cues” and the relationships of pieces to those “cues” that is at the root of my improvement.

    Obviously, I can’t “prove” any of that.

    ReplyDelete
  4. This comment has been removed by the author.

    ReplyDelete
  5. A seemingly unrelated (you decide) note:

    Last May (2018), I asked my wife to teach me how to crochet. She has only made scarves and baby blankets. After a couple of frustrating "lessons" (in which she went at her usual pace as she tried to demonstrate a couple of basic stitches), I turned to YouTube to learn the basics. My goal was to learn how to make amigurumi (Japanese: "crocheted stuffed toy") for my special needs granddaughter. I started by learning the 6 basic stitches: slip, chain, single crochet, half-double crochet, double crochet, and triple crochet. I didn't just get to the point where I could make one of those stitches when required by following along with a video or a pattern. I practiced them until I could almost do them without looking at the work. It didn't require that much time to reach that level of skill; it was somewhat similar to the rapid progress made when learning to drive. I then started making scarves (because my wife was available as an "expert" to critique my results). After making about 10 of them (each one consisting of thousands of stitches) and toboggans to go with them, I moved on to amigurumi. (I've made over 30 scarves in support of my wife's project to assist homeless people.) I got an excellent textbook written by one of the leading experts, June Gilbank from England. I started with a relatively simple animal, using it to learn how to make the various shapes of the different parts (head, body, arms, legs, tails). I made cats initially, because that's what my granddaughter loves. Since then, I have been making all kinds of amigurumi in different sizes: pigs, cats, giraffes, ram sheep (with large curved horns) plus Christmas characters (snowmen) and a variety of different Christmas figures. I couldn't find a ram sheep pattern with horns that looked realistic, so I designed my own pattern. My most recent amigurumi was Eeyore (from Winnie the Pooh). I had a pattern but didn't like the look of it, so I created a different design. I've reached the level of skill where I can look at a picture of an amigurumi (with no pattern) and make my own version of it as I crochet the various components. My wife is astounded that I can create designs on the fly, with no pattern.

    How can that be done?!?

    I KNOW (cold) the basics of crocheting, not because I read about them in a book and did them a few times by following a pattern, but because I PRACTICED the basic stitches over and over and over. (My wife would probably characterize it as being "monomaniacal".) I no longer even consider how to make one of those stitches; "it" (System 1) just does it whenever I need one. My full conscious attention (System 2) is focused on the "picture" (which sometimes only exists in my imagination) of what I want to create, rather than being focused on how to do what is necessary to create what I "see" as the finished product.

    I think this is the level of skill that we pursue in chess.

    ReplyDelete
  6. what is complicated for us is easy for a master, whats easy for us is complicated for a beginner in chess. What we try to do is: to make formerly complicated positions easy. The funny thing is: we did it before , many positions are easy for us now. So: what are easy positions for us and how did they become easy? We did see retalted positions several times before and did analysed them ( during a game..) often in different variations, so we know what to lokk for and what not to look for.

    ReplyDelete
  7. It has been said that "System 1 proposes; System 2 disposes." This is not entirely true.

    System 1 instantaneously assesses a given situation and proposes a "solution." Actually, it proposes a narrative that is internally consistent with all other relevant "facts" stored in associative [long-term] memory as experiences. An integral component of System 1 is the neglect of ambiguity and the suppression of doubt. Here's an example:

    "Ann slowly approached the bank."

    What interpretation came immediately into your conscious mind?

    Most people in industrialized cultures will instantaneously construct a mental image of a woman slowly approaching the entrance to a financial institution. However, the sentence is ambiguous. In the absence of an explicit context, System 1 generated a likely context on its own. You were not aware [via System 2] of the choice of context nor even the possibility of an alternative plausible explanation.

    Suppose we add in a contextual statement preceding the original sentence:

    "They were floating gently down the river. Ann slowly approached the bank."

    The additional contextual information guides the System 1 interpretation in a totally different direction - you can "bank on it."

    There is nothing alarming or puzzling or "out of the ordinary" about either possible scenario, so the attention of System 2 will not be activated in either case. System 2 is "effort averse" or simply lazy, and will generally accept System 1's interpretation without question as long as it does not evoke a feeling of danger or surprise.

    I think of the "vulture's eye view" as a mechanism for establishing context for a given position/problem. For example, determining what a puzzle is about (mate, gain of material, Pawn promotion) helps establish the context for finding a solution. There is also a pruning effect that occurs behind the scenes, using the "method of elimination." The effect is to narrow the range of potential solutions, while also priming the attention toward specific solutions. The vulture's eye view is NOT synonymous with Weteschnik's "status examination." In Weteschnik's process, the status examination occurs only occasionally (perhaps a few times per game), after intuition (System 1) has failed to narrow down the possibilities. It is a step-by-step logical process [System 2] of examining each and every piece on the board and its potentialities in that specific position (context) in conjunction with all other pieces. In short, it's a detailed logical cataloging of the PoPs, LoAs, and Funs of each piece. This is NOT System 1 at work!

    So, one of the fundamental training questions is:

    How can we [System 2] train System 1 to produce reliable intuitions which get us to the heart of a position?

    The traditional method is practice, practice, PRACTICE. Yet, experience has proven conclusively that mere repetition is insufficient to achieve the desired result. Anybody here with experience working in the "salt mines of Krakau" (otherwise known as the MdlM Seven Circles or the Woodpecker Method) knows first-hand the insufficiency of that method.

    ReplyDelete
  8. PART II:

    REPETITION is absolutely necessary for building intuitive skills – but is NOT sufficient.

    Consider the method of touch typing. Drills are performed until the typist reaches the desired speed with a specified minimum number of errors. The skill to be developed is the capability to look at the words in a document (without being consciously aware of what is being read) and hit the corresponding keys without consciously thinking about the key strokes required. The System 1 mechanism does the typing in response to the words that it sees; System 2 is minimally involved after the requisite skill has been established. The same kind of process occurs when learning to drive a car. Initially, System 2 must laboriously (and slowly) follow step-by-step instructions, repeating the basic steps over and over until System 1 has automagically integrated the relevant movements to produce the skill of driving – without System 2 being involved, unless a surprising or dangerous situation is detected.

    I love to draw. Do you know what I mean by that statement (the context)? Does it refer to using pencil or pen or brush to create a picture? Or does it refer to creating parallels between totally different disciplines through analogies? I do both processes.

    I love to draw pictures, especially with my granddaughter. I study books on drawing, as much for gaining knowledge as for helping her to draw more realistically. As I was studying a recent book How to See, How to Draw by Claudia Nice, I “drew” an analogy to chess. [I have a built-in “trigger” for the word “seeing”. It is ambiguous, which allows us to talk about “seeing” (as observing – sight, or as understanding - insight). Only the context determines which of those meanings is applicable.] The vulture’s eye view is another form of “seeing” which ideally combines both meanings.

    ReplyDelete
  9. PART III:

    Chapter 2 is titled “Seeing Past Preconceived Ideas”. Here are the introductory paragraphs. Keep in mind that this is an artist training book for the context.

    Observation and practice [repetition] are the keys that enable a person to draw well. Although one may be born with a patient, inquisitive nature, the SKILLS of observation may be developed by almost anyone. However, there is a major stumbling block that stands in the way of the developing artist: PRECONCEIVED IDEAS. We learn basic shapes as babies. Mother’s face is an oval. She has two orb-shaped eyes that focus on us, giving us her attention. Her smile, which we see as an upward curving bow, expresses her approval. It’s no mystery where the “happy face” symbol comes from. It is one of our first preconceived ideas of what a friendly human face should look like, in its most primitive form.

    Symbols are simple shapes used to represent an object or idea. They correspond to the preconceived ideas that are basic to most of us and are readily recognized. Think about the symbols used on road signs and on warning labels. Preconceived ideas are very apparent in the early drawing attempts of children. They don’t study the subject they are drawing, but rely on simple symbolic shapes to express themselves. When a young child draws a house, it doesn’t matter what the house actually looks like; it will most likely be portrayed as an irregular square with a triangular roof, a big rectangular door and a few box shaped windows. If there is a chimney, it will probably be sticking out of the roof at an angle. This doesn’t mean that the child SEES the house in this manner; it simply means that symbols are safer and easier to put down on paper than reality. It’s encouraging to know that even young children can be taught to be observant. As they develop their observation skills, they begin to transfer what they SEE into their art, replacing symbolic representation.

    Overcoming preconceived ideas is an ongoing process. It involves studying the actual shape, size and position of the subject you are drawing and making continuous comparisons. It means SEEING the way in which the light and shadows play over the surface of the subject and how well defined the edges are. Textures need to be registered in the sensory portion of the mind, and if possible, actually touched. Color is an important observation too, even for those working in tones of gray, because changes in hue and intensity can be represented by changes in value.

    ReplyDelete
  10. PART III:
    She gives four illustrative drawings. The first drawing is of a house by a child (or beginning adult artist), and is based almost entirely on preconceived ideas (shapes). The second drawing captures the baisc shape of the overall house but ignores tricky portions such as a side wal, where the front of the house joined the back; it was simply ignored. The third drawing is much more realistic, but the artist ignored lines of perspective along the side wall and the roof. Some comparative work with a straight edge would have helped the artist to SEE these areas more clearly. The fourth drawing is by Ms. Nice, using a lot of comparisons. Making numerous comparisons between the shapes, lines and angles of the subject and those of the drawing enables the artist to SEE what is happening more clearly. PRECONCEIVED IDEAS ARE KEPT TO A MINIMUM.

    I used this idea of SEEING (observing carefully) to assist my graddaughter to draw more realistically. I drew a typical (for her) hand using a rough square and 5 triangles. I then asked her to modify the drawing to make it look more realistic. She promptly round the ends of the fingers and made an oval of the hand AND, most importantly, she realized why her drawings of hand had seemed so unrealistic previously. One small step, but (I think) an important one. She is now able to SEE hands without relying exclusively on peconceived ideas, i.e. shapes. Does she still revert back to basic shapes? In a word, yes. One small step at a time. . .

    I wish I could assume that everyone could SEE the analogy to playing chess. Unfortunately, that does not seem to be the case. (Or, maybe I’m just “crazy” enough to think there several useful parallels. Oh well.)

    Now for the chess analogy. I think (totally unproven) that we as adults have a similar SEEING (observation) problem when looking at chess positions. We have internalized the basic “shapes” (such as forks, pins, etc.) as preconceived ideas, and we approach the problem of “drawing” a solution on the chessboard in terms of those basic shapes. In other words, we try to solve problems by applying those basic themes in those terms. If this is true, then it is not surprising that we “draw” at a very rudimentary level. This may be part of why we can “solve” thousands of puzzles with scarcely any improvement beyond the elementary level. Dr. Kahneman coined a phrase which is apropos: WYSIATI (What you SEE is all there is). In the context of chess, if all you can SEE are basic shapes (preconceived ideas), then those basic shapes are literally all you can SEE. In order to improve our skill, we have to learn how to SEE what is really there, and not rely on our preconceived ideas.

    ReplyDelete
  11. Robert said : I wish I could assume that everyone could SEE the analogy to playing chess.
    Reminds me to a training method to come back to the same positions or the same type of pattern and to try to see each time more in them.
    Maybe the danger in (wrong?) repetitions is: to see each repetition less??

    ReplyDelete
  12. Aox wrote:

    "Maybe the danger in (wrong?) repetitions is: to see each repetition less??"

    I think that is exactly why we very rapidly lose interest when repeating the same problems. We SEE less and less each repetition, and consequently fail to grow our capacity to develop the "secrets" (nuances) of chess skill. I personally do not think there is very much to be gained from repeating the same problems over and over again, especially after the problem solution (main variation only) has been memorized. A memorized problem and solution provide one very specific instance, which is highly likely to never occur in our own games. System 1 may give us a feeling of familiarity in a position that is similar, but System 1 is notorious for providing intuitive solutions that are simply wrong. Unless System 2 is monitoring and evaluating the proposed solution/narrative, we are simply unaware when System 1 proposes an answer to a simpler problem that is not applicable.

    Per Dr. Kahneman:

    The great comedian Danny Kaye had a line that has stayed with me since my adolescence. Speaking of a woman he dislikes, he says, "Her favorite position is beside herself, and her favorite sport is jumping to conclusions."

    Here are three rather simple problems from Dr. Kahneman's book, referred to collectively as Shane Frederick's Cognitive Reflection Test. The instructions are simple: do NOT try to "solve" them but listen solely to your intuition.

    Problem 1:

    A bat and ball cost $1.10.
    The bat costs one dollar more than the ball.
    How much does the ball cost?

    Problem 2:

    If it takes 5 machines 5 minutes to make 5 widgets, how long would it take 100 machines to make 100 widgets?

    100 minutes or 5 minutes

    Problem 3:

    In a lake, there is a patch of lily pads. Every day, the patch doubles in size. If it takes 48 days for the patch to cover the entire lake, how long would it take for the patch to cover half the lake?

    24 days or 47 days

    Dr. Kahneman:

    All of the problems were chosen because they evoke an immediate intuitive answer that is incorrect. . . . The experimenters recruited 40 Princeton students to take the CRT. Half of them saw the puzzles in a small font in washed-out gray print. The puzzles were legible, but the font induced cognitive strain. The results tell a clear story: 90% of the students who saw the CRT in normal font made at least one mistake in the test, but the proportion dropped to 35% when the font was barely legible. You read that correctly: performance was better with the bad font. Cognitive strain, whatever its source, mobilizes System 2, which is more likely to reject the intuitive answer suggested by System 1.

    Food for thought regarding our capacity to rely on System 1 exclusively and the likelihood that we can always trust System 1 to provide the solutions we seek. . . As former President Reagan was famous for saying (in a totally different context): TRUST [System 1] BUT VERIFY [using System 2]!

    ReplyDelete
  13. I’ve started a slightly different training regimen. I’m still “looking for” the LoA and PoPs and Funs, but I’m working through all of the Chess Tempo problems I’ve previously missed before starting a blitz session for rating purposes. It does two things: (1) I start off repeating (easier?) problems I’ve previously seen; and (2) it “primes” my recognition with “familiar” positions. Time and experience will tell if this helps. My number of “missed problems” is not that great, since my number of problems attempted (623) is not that high (yet); here are my blitz statistics so far:

    Chess Tempo Stats for blitz tactics
    Rating: 1762.2 (RD: 43.72) (Best Active Rating: 1773 Worst Active Rating: 1443)
    Active Rank: 761/2379 (Better than: 68.04% Best Active: 696 Worst Active: 1814)
    Problems Done: 623 (Correct: 496 Failed: 127)
    Percentage correct: 79.61%
    Average recent per problem time spent 41 seconds
    FIDE Estimated Rating based on blitz tactics: 1911

    That FIDE Estimate Rating is highly overrated, IMHO.


    Leonardo da Vinci:

    There are three classes of people:

    1. those who SEE,
    2. those who SEE when they are shown,
    3. those who do not SEE.


    Confucius:

    I HEAR and I forget.
    I SEE and I remember.
    I DO and I understand.

    ReplyDelete
  14. It might be , that you solve the puzzles to fast. You dont get extra points in the ct-rating when you are faster than average. But if i remember correct, you get a higher fide estimate by solving faster. To get the best ct-rating ( with the same skill ) you need to try to make relativly few reds ( missed ) and accept more yellow ( beeing to slow ) in your ct statistic.

    I repeat all puzzles where i lost more than 1.5 points, to prevent seeing less, i repeat them with extreme spacing ( min after 8 days, ...), the speed improvement per repetition is kept small . So far i have 3000 puzzles in my repetition program , almost 1 1/2 year of training.. seemingly? no effect? in my tactical skill. I will look in more detail into my statistics at end of this year.

    ReplyDelete
  15. @ Aox:

    Thanks for the information!

    There's also a penalty for getting the first move correct, and then taking some time for the remaining move(s), if any.

    Question: how do you get the display of the reds (missed) and yellows (being too slow) information? I can't find that info in the User Guide. In short, I'm not seeing that information so far.

    ReplyDelete
  16. you can see your statistic at "My Stats" -> "Tatics Stats"
    see menu on top of your page at chesstempo.
    That its better to be slow instead of missing to many puzzles is because of the "wrong" rating calculation formula.
    Empirical rabit did a lot of work about the right formula: for example here : http://empiricalrabbit.blogspot.com/2012/10/more-rating-time-and-score.html
    Richard at chesstempo chose a different one :
    https://chesstempo.com/user-guide/en/tacticRatingSystem.html#blitzRating

    In richard formula are ideas integrated to discourage guessing (= solving (too) fast)

    The formula for the fide estimate was changed several times as far as i remember. Richard did try to find a good formula
    based on statistics with his datas ( people can tell chesstempo their fide rating at one place )

    One reason why you need to solve many puzzles ( 1000-4000?) to reach your "real" rating is, that you ned to develop the right balance between speed and precision

    ReplyDelete
  17. At the given Chess tempo link, I found this explanation:

    To discourage guessing, and incremental solving, blitz problems also take account of time taken after the first move and ads an extra penalty for time taken after the first move. Essentially time taken after the first move is counted twice when accounting for total time taken. So a solve time of 10 seconds where 5 of those seconds were on the moves after the first, the total time used to calculate your rating adjustment would be 15 seconds (10 seconds + 5 seconds).

    My previous approach was to try to "see" at least the first move of the solution and put it on the board, and then try to work out the rest of the moves, since I was then one step closer to the total solution. Obviously, this is a very poor strategy given the time penalty applied to slower 2nd (and beyond) moves in the solution. That also explains why I would gain very little from many of the problems, even though I got it correct and within the time period allowed. My strategy is now changed to look for the entire solution FIRST, and then play it rapidly. THANKS!

    On a different front:

    FEN: 3rk1r1/3b1p1p/p2BpQ2/8/2q5/2P5/PP3PPP/3R2K1 b - - 0 1

    I watched IM Kostya ("IM hellokostya") Kavutskiy streaming on lichess.org a few evenings recently. He was analyzing exercise positions from The Art of Sacrifice in Chess, 21st Century Edition by Rudolph Spielmann, revised and enlarged by GM Karsten Müller. (My original Dover edition does not have the exercises that were used.) In the position above, Black is to move and DRAW.

    As I was listening to various move suggestions from the chat and from IM Kavutskiy, I realized that they were all just trying various obvious moves without a conceptual framework of what had to be done. Trial and error at its finest: throw something against the wall and see if it sticks!

    The first move is obvious: 1. … Rxg2+. After that, it gets somewhat murky. There is the immediate checkmate threat from the White Queen and Bishop, and a secondary (more distant) checkmate threat from the White Queen and Rook.

    ReplyDelete
    Replies
    1. Trial and error at its finest: throw something against the wall and see if it sticks!

      That's what I'm trying to say. Trial and error is the nec plus ultra method for system I. Concocting any kind of thought process will either add redundancy or inspires system II to kick in.

      We must analyze our trial and error and find out what hinders system I to bring it to a good end in stead.

      Delete
  18. There IS the tree of variations in which you try to find "the right" one. For trees there are different search algorithms, Depth-first and so on. Stystematical search is beneficial to random search by preventing analysing the same over and over again but with a good memory for already visited knots you can do virtually everything. Frameworks help to learn things. Its like searching for a big red spot on a white wall. As long as you have problems with it you start at the left top and.... Later you just see it. Just looking at the bottom right and bottom left corner does not help the beginner

    ReplyDelete
  19. PART I:

    Quoting GM Anatoly Lein:

    "I don't think like a tree; do you?"

    Brute force search algorithms (and the various modifications that improve the efficiency of the search process, such as minimax, Alpha-Beta cutoffs, etc.) work fine for computers, which have the speed capacity and sufficient memory to hold whatever is found in what essentially is a trial and error sequential approach to an exhaustive search of the available solution space. Unfortunately for humans, that approach requires tremendous resources that simply aren't available to us.

    Dr. Lasker recognized this problem a long time before the advent of computers.

    When looking at the results of analysis [computer or human analysis], it is true, I cannot determine by logical deduction through which particular process of thought the results have been arrived at. But to this end, though logic fail me, psychology will aid me. A spirit [computer?] with a large and roomy brain who without error could keep in mind millions of variations would have no need of planning. Frail, weak man can clearly keep in mind only half a dozen variations since he has but little time to spare for Chess. And if he by chance had more time for it and in addition had genius for the game, to see through hundreds of variations would turn his brain.

    In the example of a search for a big red spot on the wall, presume that the searcher has been told the goal - to find a BIG RED spot on the WALL. Prior knowledge is required that we take for granted and simply gloss over (or ignore). We have been given a discriminating color, its size and potential shape, and the search space is limited to a particular wall in this particular place. That reduces the search space enormously - and we are not consciously aware of that reduction by prior knowledge.

    Imagine our wall being approximately 8 feet by 20 feet. Our hypothetical searcher gets out his magnifying glass and confines his inspection to a space that will fit his lens, which we will assume is only 1 inch across. He begins his inspection through trial and error: he just starts examining specific locations on the wall at random. We will assume that he (somehow) has the capability to remember where he has searched previously, and is thus able to avoid redundant searches of the same space. Would you or anyone you know take such an approach which is NOT guaranteed to successfully find that big red spot? (For example, if the spot is much larger than the focus of the lens, it may be possible to "see red" without being able to see a "spot.")

    Perhaps there is only one red spot; if so, the searcher can immediately be sure he has reached the goal as soon as "he SEES red." Perhaps there are many red spots on the wall, of varying sizes. If so, the searcher now has to evaluate which of those spots fits the criteria of "big".

    ReplyDelete
  20. PART II:

    If you were a "coach" who was restricted to only making suggestions regarding the search process to be used and nothing else, what would you suggest to our searcher as a more efficient method?

    I personally would suggest that the searcher move away from the wall until as much of it as possible can be seen at a glance. (This would be the "vulture's eye view.")

    One of our problems is that we struggle to learn effective methods of how to teach ourselves. Like Frank Sinatra, we want to "do it MY way." So we ignore (or are ignorant of) the literature on effective learning/teaching methods. This is true whether we look solely at chess literature or peruse the more widely applicable educational literature. It's not that we consider it and then reject it. Instead, those potentially applicable methods are outside the scope of our focus on chess improvement (our 1-inch lens, if you will).

    Searching the same old method space in the same old way using the same old tools AND expecting different results - is the very definition of insanity. If a method does not seem to be effectively working, perhaps looking in a different search space for a method MIGHT prove to be more effective.

    Or maybe I'm just "crazy."

    ReplyDelete
  21. well i did try different methods.. i am waiting for a "new" one

    ReplyDelete
  22. Perhaps an analogy will make my point clearer. Many millions of manhours have been expended on improving (incrementally) computer systems based on the von Neumann architecture. Many more millions of hours have been expended on implementing and improving Claude Shannon's Type A search strategy for chess. The only known limitation is apparently the speed of the underlying hardware.

    Then along comes Alpha Zero, with completely different hardware and a completely different approach and it defeats the reigning champion of the previous paradigm. It also has no apparent (at this time) limitation on its capacity to improve, given the possibility of expanding the hardware in parallel, as well as making faster hardware. Or take Leela Zero, which is a software implementation on conventional von Neumann hardware.

    If no one had considered changing the method and looking outside the current paradigm, we would still be focused on trying to find miniscule improvements within the same search space.

    In the past, my approach to chess improvement was essentially trial and error. I'd try some idea suggested by a "guru" and (maybe) experience some small improvement in a very specific area (that "little tiny red spot"). My current approach is based on the discussions on this blog and a knowledge of various educational methods based on the deliberate practice paradigm. I'm trying to look at "the wall" FIRST. I'm still working out the kinks and trying to be consistent in my practice using that method. So far, the results have been encouraging.

    Will I eventually reach a plateau? I presume that likelihood has a probability of 1. Then I will have to find a different approach that will allow additional improvement. Lather, rinse, repeat, ad nauseum, but don't "flog a dead horse." The point is NOT to find an "El Dorado" method, but to find something (anything) that efficiently increases my skill level over time.

    ReplyDelete
  23. PART I:

    Here's another example of how we don't see the method forest because of the red spots in our eyes from massed practice.

    make it stick - The Science of successful Learning - Brown, Roedinger III, McDaniel

    Interleaving practice of two or more subjects or SKILLS is also a more potent alternative to massed practice, and here's a quick example of that. Two groups of college students were taught how to find the volume of four obscure geometric solids (wedge, spheroid, spherical cone, and half cone). One group then worked a set of practice problems that were clustered by problem type (practice four problems for computing the volume of a wedge, then four problems for a spheroid, etc.) The other group worked the same practice problems, but the sequence was mixed (interleaved) rather than clustered by type of problem. [Anybody "see" the parallels to training chess tactics by tactical device/theme?] Given what we've already presented, the results may not surprise you. During practice, the students who worked the problems in clusters (that is, massed) averaged 89 percent correct, compared to only 60 percent for those who worked the problems in a mixed sequence. [AHA! We've found that massed practice ala MdlM or Woodpecker Method IS the way to go for long-term chess improvement! All we need to do is modify the method a little so that we work on one specific tactical device/theme until we have it down "cold." Not so fast. . .] But in the final test a week later [trying to apply what has been "learned" in a tournament game] the students who had practiced solving problems clustered by type averaged only 20 percent correct, while the students whose practice was interleaved averaged 63 percent. [RUH-ROH! The second group RAISED their scores a week later!] The mixing of problem types, which boosted final test performance by a remarkable 215 percent, actually IMPEDED performance during initial training.

    . . .

    The learning from interleaved practice FEELS slower than learning from massed practice. Teachers and students sense the difference. They can SEE that their grasp of each element is coming more slowly, and the compensating long-term advantage is not apparent to them. As a result, interleaving is unpopular and seldom used. Teachers dislike it because it feels sluggish. Students find it confusing: they're just starting to get a handle on new material [that System 1 feeling of "familiarity"] and don't feel on top of it yet when they are forced to switch. But the research shows unequivocally that mastery and long-term retention are much better if you interleave practice than if you mass it.


    The referenced book is NOT a "chess" book per se. Looking outside the chess literature is what I referred to previously regarding the search for workable methods. How many books have you read that are NOT about chess per se, and then tried to draw knowledge and analogies from them to chess which can be applied as an improved chess training method?

    ReplyDelete
    Replies
    1. "But the research shows unequivocally that mastery and long-term retention are much better if you interleave practice than if you mass it."

      My conclusion is slightly different.
      Clustering is apparently the fastest way to acquire new knowledge.
      Mixing is the best way to make it stick.
      So the two methods should be combined.

      Delete
  24. PART II:

    I've done this process throughout my entire life. I learn something in one area and then try to apply it using analogies to many different areas. Two of my favorite philosophical influences are Miyamoto Musashi and Dr. Emanuel Lasker.

    Miyamoto Musashi: [GO RIN NO SHO – The Book of Five Rings]

    These things cannot be explained in detail. FROM ONE THING, KNOW TEN THOUSAND THINGS. When you attain the Way of strategy there will not be one thing you cannot SEE. You must study hard.

    Dr. Emanuel Lasker: [LASKER’S MANUAL OF CHESS]

    You learn no art by anxiously restricting yourselves to it; YOU HAVE TO SEEK ITS ASSOCIATION, AND ITS LOGICAL CONNECTIONS AND ANALOGIES WITH THE REST OF THINGS. Otherwise, you will learn no more than the craft, the technique of your art and never attain to a full comprehension or easy mastery of it.

    [I'm reminded of Kenny Werner's excellent little jazz training book, titled "Effortless Mastery"; there is no jazz music in it.]

    I want to train my pupils to think for themselves and exercise just criticism. I will not teach them mere formulae, mere generalities, but will instill into them lasting principles that will grow and blossom; which are alive and vital. They must be ready and willing to put their conceptions, laws, and valuations to the proof, again and again, diligently and cheerfully, from a sheer joy of the law and from veneration of the fact.

    Philosophies to live by!

    ReplyDelete