Thursday, June 16, 2016

From complex to simple in ten minutes

Some things have gone awfully wrong during my chess education. For such flaws in my skills, I pay the price time and again. There are positions where you know you should see the clue immediately, but you don't. For instance this position:

Black to move
4N3/1p1r1p1k/p1p1r1np/2P1nRp1/1P2PbP1/2B4P/1P2BK2/3R4 b - - 1 1
solution

The last move of white was 1.Ne8. Can I take it? What worries me is the rook on d7. Its defender(s) can be harassed. Can I exchange this problem piece off with 1. ... Rxd1 first? But what about the zwischenzug 2.Nf6+ ? The knight sits well defended on f6.

After I found the solution, I spent ten minutes about thinking why this position takes me so much time, and how I could fix that. I came up with the following reasoning:

1. ... Rxd1 captures a rook. This places the obligation on whites shoulders to take a rook back. White can postpone his obligation with the threat 2.Nf6+ as a zwischenzug. This would work if black had to remove his king. But if black can get rid of the threat by capturing the attacker, and he can, by playing 2. ... Rxf6, he would solve his problem and now he has placed two obligations on the shoulders of white, who must take two pieces back. Since white can only fulfil one obligation in one move, he looses a piece. This idea is always profitable, even when you use two rooks as attackers, and the targets are two minor pieces. In which case black  would win two minor pieces for a rook.

This simple rule, in case a capture is met by a threat in stead of a recapture, is usable in a whole lot of situations. It makes those situations much less complicated to judge, without the need to calculate every single line. Finding this rule is the best well spent ten minutes in a long time!

We need to find the flaws in our chess education and fix them. It is better to study 500 positions in depth, than 200k+ positions without learning anything from them. I'm sure there are 4-5 more of these easy fixes waiting for me out there!

24 comments:

  1. you have to calculate "all" Captures and you start with the most forcing ones = taking the piece with the highest value. So this example simply shows the value of the CCT-Calculation Algorithm : calculate the ( seemingly ) most forcing moves first.
    The "best move" is the real "most forcing" move



    ReplyDelete
  2. In my opinion this position (task) is WAY harder than we perceive it. What I mean is the subpositions we should analyse, evaluate and draw conclusion every time the change is made. This way we can evaluate it much better and decrease the level of error.

    The most important is the CONCLUSION you made: try to check out the captures first EVEN if they look ridiculous at first glance! I have already mastered this concept, but ONLY at the extremally simple case: desperado! For about 100K games I have played so far... I missed this concept in less than 5% cases (positions). I do not know why I have so high score at this concept (motif), but probably because I love the idea: sacrifice the piece for the biggest price if you have to lose it anyway. It is connected with the concept of "zwischenzug" (intermezzo) and these two concepts merged together are my favourite ones. Of course I mean the scenarios when there are NOT a lot of exchanges, captures and surprising moves!

    ReplyDelete
    Replies
    1. It seems that our conclusions contradict each other. I say the position is much simpler than it looked at first sight and you say it is more complicated than it looked :D

      Delete
    2. What I understood is "you do not have to search a complex method how to solve it" while my perception is: "unless we analyse the position and all the worthy elements into details - we are not able to make a final (important and useful) conclusions".

      In general - this puzzle is complex to me unless I will analyse it into details and make a final conclusion. It does not mean it is complex for others, but it shows how weak I am at evaluating the positions complexity ;)

      Delete
    3. I'm not interested "in all the details" YET. I'm sure there are a whole bunch of details I haven't discovered yet. But that is for later, maybe.

      Somewhere, Dan Heisman described how he solved a certain position. What stood out for me, is that he pruned the tree of analysis without any hesitation, due to his ideas about CCT.

      I tried to imitate that, by taking it literally. So I started to investigate every check, every capture, and every threat. Only to find out that the majority of these CCT moves played no role at all. There are a few more rules, which he uses but forgot to tell us about, since he uses them unconsciously.

      In this post, I describe the discovery of such rule: if I capture a piece, and my opponent doesn't recapture, but poses a threat instead, he will loose wood if I am able to capture that threat.

      This is so self evident, that it needs no calculation. Before you even can think about playing chess, you need to see what is self evident as self evident. And if you missed something self evident in the past, it is high time to correct that omission. There might be intricacies in the position that make that a self evident idea doesn't work in this specific case. But I'm totally not interested in those exceptions (yet). I'm interested in learning what is self evident.

      Delete
  3. I thought it was fairly simple after a couple quick calculations. The two main capture choices were taking the knight on e8 or taking the rook on d1. The first step is seeing that these two options HAD to be looked at - I guess some people might only see one or neither of the options. After that the key is seeing the relationships of all the other pieces. I looked at taking the knight first as this piece seemed "free." I think looking at the other capture first was also an option, but I figured superficially that my rook was safe as it was protected. However, after a quick glance, I saw that 1...Rxe8 fails to 2.Rxd7 Nxd7 3.Rxf7+, which led me to taking the rook first, of course checking to see if there were any bigger threats that the knight (or rook) posed.

    ReplyDelete
  4. I agree with your conclusions that learning from a fewer number of positions is better than NOT learning from a greater number of positions. However, that is assuming "zero" learning occurs when doing those positions. For example, when I study master games (and yes, I know we're talking about tactical problems), sometimes there will be a move that I don't understand. I try to figure it out through analysis, reading the annotations, as well as asking others, but if I don't figure it out...I just keep it as a note in my file to look up in the future (if I think the position/move is important enough to do so). More often than not, after studying similar games in the same opening the reasons for the move become apparent, because I saw it in slightly different contexts perhaps, or because in the other game the opponent took advantage of the ommission of the move or whatever. The point is, that sometimes the "learning" takes place over time and experientially as well as "deliberately." I'm not saying one way is better than the other, but I think there is a value in quantity as well as quality - and perhaps the answer is a balance of the two. I suppose this is also a question of how consciously we can or should examine this process as opposed to developing experience and skill through repetition (with feedback) of the exercises. In any case, some food for thought.

    ReplyDelete
    Replies
    1. I have done 120K+ problems in the past, and I can guarantee you that it is entirely possible to learn nothing from them. I assume it is possibly to exaggerate in the other direction as well, and going too deep. But that can only happen when you study a rare position. In that case you waste your energy. But positions like in this post are extremely common, and I noticed I had trouble with it, and in this case there is no way I can go too deep here. Since I MUST learn to see it as simple as it is, and I MUST take away the cause of my confusion. Since there is no reason to be confused.

      A measly 500 puzzles, is that a lot or not a lot? I'm pretty sure that those 500 puzzles will be good for demonstrating 95% of the biggest flaws in my thinking, and if I would be able to fix those flaws, I would make a quantum leap forwards.

      A balanced approach sounds always wise. But if you want to run the marathon, and you are exhausted after 3 minutes, does it make sense to worry about your walking technique? You have to work on your physical condition first.

      Delete
    2. Well, think about shooting a basketball. If your technique is bad, you won't make many shots. however, if you just correct your technique but don't put in enough reps, you won't progress very much either. you have to do both. Maybe this isn't a point you are arguing.

      Delete
    3. I don't have opinions. I just use myself as a guinea pig to test hypotheses. The more extreme the test is, the clearer the conclusion will be. With a balanced approach you don't know which factor is paramount. I have tried an extreme balanced approach in the past too, btw :D

      Delete
    4. I have done 400 000++ puzzles without significant improvement. I watched several teenagers improving for several years though. But i found no adults ( >30 ) doing so.

      Delete
    5. [Quote]"A measly 500 puzzles, is that a lot or not a lot? I'm pretty sure that those 500 puzzles will be good for demonstrating 95% of the biggest flaws in my thinking, and if I would be able to fix those flaws, I would make a quantum leap forwards."

      I fully agree! If you are able to present 500 puzzles (positions) you are making these wrong - after you fix the mistakes you are doing while solving - the chance you will make progress (at tactics) is very high.

      When I have been doing 555 chess puzzles related to the tactical opening mistakes - I realized there are very similar ideas merged together. And after some repetitions I discovered without tactical flaw (weakness) you cannot win material (in a sound way). It means - pattern and motifs recognition maybe a key to fasten the process of finding the final (best) solution.

      That's why I scored 2372 rating at ELO meter [http://www.elometer.net/] test. And guess why? It is because the majority of tasks (about 80%) were very close to the ones I memorize (remember) the most. Otherwise I could score no better than 1850-1900 (as I did in the past with the positions I was not familiar).

      Delete
  5. According to Chess Tactics Server (CTS), I've "tried" 34824 problems, with a success rate of 78.9%. I've also done problems on Chess Tempo and Chess Academy and other places and in a lot of chess tactics books, but nowhere near as many as CTS. Given that my current (last) rating on CTS is a mere 1508, I have to conclude that it was pretty much a waste of time. The rather funny (not "Ha ha!") thing is that my Chess Academy rating has stayed above 1800, bouncing up and down to about 2000. The critical difference: Chess Academy does NOT time how long the problem takes to solve. I have noticed that I usually take approximately 15-45 seconds for most problems, with some outliers that take several minutes. It would be severely discouraging IF I was looking to get a high rating at tactics problem solving.

    I just finished Daniel Kahneman's thought-provoking book, Thinking, Fast and Slow. He divides thinking into two systems, named (appropriately) System 1 and System 2. System 1 is the intuitive, massively parallel processing sub conscious "thinking" which uses long-term memory. System 2 is the logical, sequential processing conscious "thinking" which uses short-term memory.

    One of his thought-provoking statements is this: System 1 is not readily educable. It is not directly accessible nor directly trainable, but it is the repository of expert skills.

    That raises the question: why would the subconscious be the repository of expert skills, yet not be readily (or directly) educable (trainable)? Obviously, youngsters do not seem to have this problem of training, and seemingly have more direct access to System 1. In contrast, we adult "improvers" (we wish!) base our approach on conscious and logical thinking (System 2), for which we have the greatest affinity in our training methods - and it shows in our results.

    Perhaps we need to rethink our training methods and move away from logical conscious processes as far as possible, and try more primitive (NOT primitive positions!) processes that are not so logical nor conscious, in fact, that are completely non-verbal and subconscious. I have some ideas in mind, but unfortunately have insufficient time to explore them in the form of one or more computer programs.

    One of those ideas is to extract the position of the piece(s) that should be observed FIRST in a given position, change the degree of emphasis of those specific pieces (perhaps shifting to a darker color) for a very short period of time (much less than a second) so that the change of emphasis is subliminal, repeat this shifting for a very short time period (less than 5 seconds) and then query/test the "student" about the important pieces in the position using point and click. This is all done visually, not using words. The results are stored in the computer just as for current tactical training, and spaced repetition is applied based on the results. There would be no "principles," no list of variations, none of the verbal conscious stuff that we normally use to "discuss" (internally or externally) a "word picture" (description of what is going on in that position). Obviously, this can be done in several different ways with many different concepts at different levels. The basic idea is to try to directly interact with System 1 without using System 2 (or using System 2 as minimally as possible).

    This would require either an enormous amount of work in order to determine effective positions for training, or it would require an advanced "problem generator" program that could extract critial typical training positions from a chess game database and automatically identify the "significant" piece(s) in that position.

    Blue skies. Smiling at me. Nothing but blue skies. Do I see. . .

    ReplyDelete
    Replies
    1. Robert said: "Perhaps we need to rethink our training methods and move away from logical conscious processes as far as possible, and try more primitive (NOT primitive positions!) processes that are not so logical nor conscious, in fact, that are completely non-verbal and subconscious."

      That is what we tryed with the board vision training. We did improve in improvable ( extreme simple / board vision ) pattern and did hope that we would become able to improve in more complex ( M1 / tactical ) pattern
      Our succcess was not overwhelming

      Interesting that Daniel Kahneman's did come to the same results as me. As an adult unimprovable high speed processes based on patternrecognition and an improvable k = slow process.

      I suspect the reason for the difference of adults and kids is based on the development of the brain.
      But Daniel Kahneman ( and i ) might be wrong. Maybe there is some trick

      O_o

      Delete
    2. Given my previous training, I suspect that the "trick" that is used by kids (subconsciously) is referred to in Zen circles as "beginner's mind." We lose that simple "viewing" capability as we mature into adulthood. It is extremely hard to let go of the conscious control of our thoughts once we have acquired at least some control over them. This prevents us from "seeing" what should be obvious. Instead, we intellectualize everything using System 2, thereby blocking System 1 from operating efficiently and spontaneously. Your famous example of the "invisible gorilla" is a perfect illustration of the blocking of System 1's "view" because System 2 has been explicitly instructed to ignore everything (and that includes anything System 1 might advise, even though we are unaware that we are blocking it) except counting the number of passes by the white team. Kahneman discusses this "blinding" process using this very example!

      The classical example of acquiring "beginner's mind" is through the use of the Zen koan. It is completely baffling to the logical mind, and can be extremely frustrating to logically minded and highly intelligent people. The harder you try to work it out logically, the farther away from the aolution you get. Yet the "seeing" (directly experiencing) mind (System 1) can readily comprehend the "solution" - once the logical mind (System 2) is completely blocked out.

      The most famous Zen koan is this: "What is the sound of one hand clapping?" If you understand that koan, then you can (in theory) and should "see" the solution to any and all Zen koans. My one hand claps its approval. . .

      The "problem" we are trying to solve as adult chess improvers is very similar to solving a Zen koan. Once we "see" the solution, it becomes impossible to "not see" it. I liken it to your excellent example of "seeing" four-leaf clovers. The same thing occurred with Temposchlucker's grandmaster level of "seeing" a Knight move around the board. The specific skill, once acquired, lasts forever. Part of the problem we face in chess is that there are a considerable number of very specific skills that must be acquired and then coordinated/correlated/coagulated using both System 1 and System 2 in order to actually play chess at a high level of skill. This is so much easier said than done!!

      Delete
  6. I'm not so far yet. This blog is aiming to find an educational method for system I since 2005. In this post I try the following: to see the solution in stead of calculating it. In this specific case, it took me a ten minutes AFTER I found the best solution with verbal reasoning to actually SEE it. But it is not always so simple. Since yesterday, I'm busy with a new position where I find it difficult to SEE the solution. Calculate it, no problem. But SEEING it, is an entirely different animal. I know I can do it, but I'm definitely out of my comfort zone. I seek distraction, check my e-mail, read facebook, write a comment on my blog, all to PREVENT me from learning to SEE the solution. But I will persevere, until I finally SEE the solution without the need to calculate. So it STARTS WITH LOGIC, and usually it ends there as well. But it must be translated to SEEING IT WITHOUT CALCULATION OR VERBAL REASONING.

    SHOOT, I wrote a comment in stead of trying to SEE the solution of this new position AGAIN!

    ReplyDelete
    Replies
    1. In school some schoolmates and i had a battle in finding the most four-leaf clover. For many months we spend our breaks to beat each other in finding as many four-leaf clover as possible.
      Even today i SEE four-leaf clover in a place with clover, i dont need to search it, it pops in my attention more or less by itself. Of coure sometimes these are just 2 clovers close together but i would beat any "normal" person in searching them by far!

      Delete
  7. Some of what I found fascinating is that System 1 can provide "suggestions" almost instantaneously, BUT (there's ALWAYS a "butt" in there) it is just as likely to provide the answer to a somewhat different "simpler" question than to the actual question asked IF there is sufficient similarity between the two questions - and we are NOT consciously aware of the substitution! (I think this is where we go off the rails initially when trying to orient ourselves in a position, not realizing the "answer" that came immediately to mind is NOT the answer to the question originally being asked. That is one significant aspect of training the vulture's eye view.) System 1 interacts with and relies on System 2 to sort out the wheat from the chaff BUT (another "butt" in the way of a clear-cut answer) System 2 is "indolent." (I would have used the word "lazy" [Google defines "indolent" as "wanting to avoid activity or exertion; lazy"] but I'm not a PhD Nobel Prize winning psychologist.) System 2 does NOT want the responsibility for, nor the effort involved in, making a good decision (see your comment about the process of avoidance above), so (in far too many cases of chess decisions) it essentially just "goes along to get along" without any serious effort to figure out the possible downside of the immediate System 1 suggestion. I think (unproven assertion) that this is one area in which we amateurs differ significantly from the masters. As an amateur, I often am willing to accept "good enough" rather than try very hard to find and verify the very BEST objective answer. That is more of a mental discipline issue, which I still work on.

    My suggestion (purely System 2) above is an attempt to bypass System 2 altogether while trying to directly influence System 1 to recognize specific patterns. I think (untested hypothesis) that it is possible to do this with chess pattern training. I postulate that if isolated and left to itself, System 1 will automatically trigger and correlate corresponding thoughts of possible tactical themes (which are already known by System 1 from the salt mines and hundreds of thousands of tactical problem-solving). I think the patterns needed are already embedded in System 1; unfortunately, we don't have a consciously controllable trigger mechanism to cause them to be recalled.

    I also think (confirmed by virtually all Knight Errants) that our current training methods have proven to be inadequate (insufficient) for the intended purpose (rapid "seeing" of the gist of a position with corresponding rating increase). We have mostly automated the manual process of working through a book of chess tactics problems, without taking advantage of the unique capabilities of the computer to do what otherwise cannot be done: virtual direct access to System 1, bypassing System 2 altogether.

    I'm well aware of how exotic this all sounds, and it is useless speculation without computer programs to test the hypothesis. Unfortunately, I don't have the programming skills in current graphical envronments nor the time to devote to developing this stuff to a working level at this time.

    Old Radio Shack ad for TRS Model I computer: "You are only limited by your imagination."

    Crazy Bob's first corollary: "Most people cannot imagine how limited their own imagination really is!"

    ReplyDelete
    Replies
    1. Robert said: "Unfortunately, I don't have the programming skills in current graphical envronments nor the time to devote to developing this stuff to a working level at this time."

      HTML is an "easy" possibility to display things and watch for reactions of the user.
      If you could give an example what and how we may think about a "workaround" for the "graphical things".


      Delete
  8. I analysed the position moving the pieces at the board.

    My short conclusions:
    1. If you can REFUTE easy one of the (candidate) moves - you can save a lot of time. You can be sure of the moves simply does not work [1... Rxe8 2. Rxd7 Nxd7 3. Rxf7+ Kg8 4. Rxd7]
    2. If you gain the material - try to force your opponent to waste time to recapture it. If he doesn't he loses material and if he does it in a not optimal way (Nf6+) make him not to have time to recapture (Rxf6!) both exchanges/captures (Tempo's discovery).
    3. Try to have a list of exchanged (captured) pieces from both sides. It may be helpful to check if you gain or lost material.

    Your idea related to the "being overwhelmed by too many obligations" is a very good one! It helps to cut off the variations we do not have to take into account (with some exceptions of course). This way we can save time for counting necessary captures.

    BTW. If you wish you can compare what variations I have tested (wanted to check out) at this position. If you have any idea which ones are irrelevant and can explain it to me - it will be great! Thanks in advance!

    -------------------------------------------------------------------------------
    [Result "*"]
    [SetUp "1"]
    [FEN "4N3/1p1r1p1k/p1p1r1np/2P1nRp1/1P2PbP1/2B4P/1P2BK2/3R4 b - - 0 1"]

    1... Rxd1

    [1... Rxe8 2. Rxd7 Nxd7 3. Rxf7+ Kg8 4. Rxd7]

    2. Bxd1

    [2. Nf6+ Rxf6 3.
    Rxf6 [3. Bxd1 Rxf5 4. exf5 Ne7] [3. Rxf4 Rxf4+ 4. Ke3 Rh1] [3. Rxg5 Bxg5+ 4. Kg2
    Rd7] [3. Rxe5 Bxe5+ 4. Ke3 Rh1] 3... Rd7]

    2... Rxe8 3. Bxe5 Nxe5 *
    -------------------------------------------------------------------------------

    ReplyDelete
    Replies
    1. I just entered Tomasz's FEN into Fritz, and found something interesting. In the original diagram given in this post, Black is at the bottom of the board. In Fritz, Black is at the top of the board. So what, you might ask? Well, when I first viewed the position, I initially "saw" the relationship between WRd1 and BRd7 as important, only noticing the potential fork Nf6+ later.

      When the board was reversed, I immediately "saw" the "threat" of 2. Nf6+, and had the "feeling" that I must not allow it because the BRd7 and BKh7 would be the targets.

      Reversing the board representation caused me to look at the position differently, even though I was already familiar with analyzing that exact position quite recently!! That felt really WEIRD, almost mind-bending!

      Maybe this is something we should actually try to do: switch the board around during the "vulture's eye view" phase (orientation) and see if something different springs to mind as a possible "solution." Fritz gives that option, but I never have used it; I certainly will now.


      Delete
    2. I hadn't checked Chess Tempo (CT) before for the "flip the board" feature, but it is there in the lower left corner of the board. On the other hand, Chess Tactics Server (CTS) and Chess Academy do NOT have this feature.

      Delete
  9. It is a known phenomena. There were experiments and tests related to the identical (exact) position with reversed colours (or views). Many people could see the position anew.

    It may be interested to check it out, but to me it is not anything that can bring significant change in the search of best solution (move). It may just be a good exercise with reversing the board when we cannot see the position clearly. I did it many times when I played against players at FICS and I wanted to see what my opponent can do to my (weak) position. Sometimes it really helped!

    ReplyDelete
  10. There seems to be one method of education for system 1, though:
    Sportsmen like racing drivers visualize the race track, gymnasts visualize the exercise they are about to begin. And that was how I learned to shift gears, by imagining it at home, sitting on a chair.

    And that is what I tried to do in this post. First logical reasoning, then visualize it. Without the logical reasoning prior to the visualization, there is nothing to visualize. System 1 is semi intelligent. Someone must keep an eye on it.

    ReplyDelete