I always have suspected that De La Maza left out some crucial information in his articles. Not on purpose, but because he didn't know it. Or didn't deem it as important.
After 15 years of investigation, I think that I know what that missing information is. Or at least a part of it.
When De La Maza did his circles way back, he did something accidentally right. He educated his system 1 (subconscious thinking).
His articles and book were based on his interpretation of what happened by his system 2 (conscious thinking). And system 2 (conscious thinking) has no clue of how to educate your system 1 (subconscious thinking). So his story was actually based on a misinterpretation of something that happened to system 1 (subconscious thinking), and which accidently yielded results.
Nobody of the Knights Errant has been able to come even remotely close to his results. That isn't a surprise, since we were guided by the concoctions of his system 2 (conscious thinking). And system 2 (conscious thinking) has clearly no clue of what works for system 1 (subconscious thinking), since they don't speak the same language.
What we found out, is that system 1 (subconscious thinking) can do something with chess diagrams and colored squares and arrows. The preliminary work can be done by system 2 (conscious thinking), but then it has to shut up and give system 1 (subconscious thinking) the time to works its miracles.
It works. Albeit I still have the feeling that the method is suboptimal. But at least we have a direction for further investigation.
Euwe said: Tactics is seeing, Strategy is thinking. Which gives us an indication of the language of system 1 (subconscious thinking): "seeing"
ReplyDeleteI like Dr. Kahneman's aphorism: WYSIATI (What You SEE Is All There Is).
ReplyDeleteAssume that the discovered System 1 “learning process” (utilizing PoPLoAFun in conjunction with diagrams, drawing circles and arrows to represent the points of pressure and lines of attack) IS suboptimal. That would hardly be surprising, given that we have no OPTIMAL empirical idea how to go about force-feeding System 1 with the required skills in the shortest amount of training time.
Voltaire: “The best is the enemy of the good.”
MdlM’s “method” is a simplification (a narrowing of focus, so to speak) of the “tried and true” suggestion of studying master games. He also recognized the limitations, noting how difficult it would be to raise his playing level using the same training method. (I'll leave aside any discussion of his supposed "cheating" using a computer up his sleeve.) Whether focusing on tactics (regardless of whether it is “thought process” or “motifs” or “tactical themes/devices” or “checkmate patterns”), or trying to inculcate those ideas indirectly by playing over myriad master games, there is so much that must happen “behind the scenes” within System 1 in order to have an increase in skill. It is ASSUMED by those advocating this training approach that extended exposure to a significant quantity of the requisite material is sufficient for the student to gain significant improvement. Alas, as already shown from the Knights Errant, that is not the case.
Additionally, we have enormous empirical evidence over centuries that playing over master games (no matter how many) is not the optimal path to chess mastery. Consider the enormous number of players who follow this training regimen - and the almost identical numbers who still fail to gain mastery or anything approaching it. There MUST be more to that training approach than meets the eye.
"There are none so blind as those who WILL not 'SEE'."
In reality, these recommended paths (including MdlM’s Seven Circles) CAN work – but only for those who know how to “read between the lines”. We do have more modern methods for trying to achieve System 1 training, increasing skill. The “deliberate practice” methodology seems to work best, coupled with spaced repetition and intense focused attention. We learn to do better by DOING, not through abstract studying and theorizing. Our focus should be on skill development, not on optimizing our newly found methodology.
My own orientation is toward practical application of theory, not development/expansion of theory qua theory. So, I'm not really interested in optimization - at this point.
Moving toward practical application:
Here are two problems from Impove Your Chess Tactics - 700 Practical Lessons & Exercises - Yakov Neishtadt, New In Chess, © 2011. The tactical theme/device given is Deflection. (Hint, HINT!) Neither problem is particularly difficult to solve, but there are interesting variations. It is instructive for System 1 to take different approaches to solving these problems. Try CCT, looking for tactical themes/devices, PoPLoAFun, whatever floats your boat, using any “thinking process” you feel comfortable with.
Problem #65:
FEN: 1R2r1k1/2pq1ppp/3p4/3P4/1Q3P2/8/r5PP/5RK1 w - - 0 1
Problem #69:
FEN: 1R2rq1k/2p3p1/Q2p1pPp/4p3/4P3/8/P1r3PP/1R4K1 w - - 0 1
Problem 69 has 3 winning first moves. It basically are 3 problems in one. However that might be seen as beautiful, it might not be the smartest way to teach your system 1 (subconscious thinking) the basic patterns.
DeleteMaster Neishtadt gives the following “solution(s)” to Problem #69:
DeleteFirstly, 1. Rxe8 Qxe8 2. Qa4! and Black cannot take the queen because of back-rank mate, whilst the counter-blow 2. … Rc1+ fails to 3. Kf2. The second method was 1. Qb5! After 1. … Rxb8 (or 1. … Rc1+ 2. Kf2+ 3. Ke1 Rc1+ 4. Kd2) 2. Qxb8 Kg8 (there is nothing else) 3. Qb3+ White wins a Rook.
Neither of these tactical operations occurred in the game Capablanca-Thomas (Hastings 1929/30). Upon 1. Qa8, Thomas resigned, but wrongly! After the deflecting counterblow 1. … Ra2! White would have to play a heavy-piece ending two pawns down, e.g.: 2. Qxa2 Rzb8; 2. Rxe8 Rxa8! 3. Rxf8+ Rxf8; 2. Qb7 c5 3. Qc7 Ra8!
GM Stockfish (after a relatively short 11-minute “look”) gives these 3 winning variations:
D35 +21.48 1.Qb5 Rxb8 2.Qxb8 Kg8 3.Qb3+ d5 4.Qxc2 Qe8 5.exd5 h5 6.a4 e4 7.Qc4 Kh8 8.d6 cxd6 9.Qf7 Qxf7 10.gxf7 Kh7 11.a5 Kg6 12.Rb7 Kg5 13.a6 g6 14.Rb5+ d5 15.Rxd5+ f5 16.a7 h4 17.f8=Q h3 18.a8=Q hxg2 19.Qe7+ Kg4 20.Kxg2
D35 +7.25 1.Rxe8 Qxe8 2.Qa4 Rxg2+ 3.Kxg2 Qxg6+ 4.Kh1 Qg4 5.Qc2 Qc8 6.a4 c5 7.a5 c4 8.Rc1 Qa6 9.Qxc4 Qxa5 10.Qc8+ Kh7 11.Qf5+ Kg8 12.Rc8+ Kf7 13.Qd7+ Kg6 14.Rc1 Qc5 15.Rg1+ Qxg1+ 16.Kxg1 h5 17.Qxd6 Kg5 18.Qe7 Kg6 19.Kf2 h4 20.Kf3 Kh7 21.Kg2 Kg6 22.Kh3 Kh6 23.Kg4
D35 +3.51 1.Qa7 Rc1+ 2.Rxc1 Rxb8 3.Rxc7 f5 4.Rf7 Qd8 5.Rxf5 d5 6.exd5 Rb1+ 7.Rf1 Rxf1+ 8.Kxf1 Qf6+ 9.Kg1 Qxg6 10.Qb8+ Kh7 11.Qxe5 Qb6+ 12.Kf1 Qa6+ 13.Kf2 Qxa2+ 14.Kf3 Qa3+ 15.Kg4 Qb4+ 16.Kf5 Qf8+ 17.Ke4 Qb4+ 18.Qd4 Qd6 19.g3 Kg8 20.Qe5 Qb4+ 21.Kf3 Qb5 22.h3 Qf1+ 23.Kg4 h5+ 24.Kxh5 Qxh3+ 25.Kg5 Qh6+ 26.Kg4 Qg6+ 27.Kf4
I wasn't concerned with the asthetics of the two problems. The problems struck me as having common elements (back-rank weakness, overloading of pieces, duplo attacks and deflection). I cannot (yet) judge the "smartest way" to teach System 1. However, I do know that SEEING common elements does favorably impact System 1 in the long term. System 1 has the marvelous ability to automagically meld common elements into stronger neural relationships, making it more likely for System 1 to bring those common elements up to conscious thought at a later time as an "intuition".
It is a very useful problem. I tried to dissect it. It struck me that I need a few hours to do so. The dissection is mainly done by system 2 (conscious thinking). The education of system 1 (subconscious thinking) starts only after the dissection is completed. That is my experience.
DeleteThere are so much more basic problems that need less preparation by system 2 which deliver patterns for system 1 to play with. It is a matter of diminishing the activity of system 2 while maximizing the exposure time for system 1.
As you know, I'm a fan of offering the problems by theme. Or motif ;) In order to maximize the chance to see analogies between positions. This position tells me, that it would be a good idea to theme the problems by material too. Problems without the minor pieces, have a certain "taste". Meaning, they have common patterns. Which are obfuscated by adding the minor pieces.
System 2 is terrorizing our training time. The cocky system 2 tends to claim all time and energy for its pedantic concoctions. The docile system 1 is assisting system 2 by fetching everything it needs for showing off its brilliancy. In doing so, system 1 is totally occupied by serving system 2, while system 2 is totally emerged in its own activities.
ReplyDeleteTo feed system 1, it needs "exposure" time to the patterns. I'm on day 94 of my training regimen. The first 52 to days were totally absorbed by the activities of system 2. Only by giving system 1 exposure time to the problems, it could work its miracles.
By "optimizing the training", I mean increasing the exposure time of system 1 to the patterns. While system 2 shuts up.
De La Maza's main accomplishment wasn't a chess understanding one, IMHO. MDLM's main accomplishment, imho, was a very impractical (but he was willing to be his own guinea pig) system for reaching peak performance. He got very messy positions, which allowed his strategy of tactics, and calculation, to work for him. He basically "crammed for an exam".
ReplyDeleteHe hit expert ONE TIME! The biggest weakness in his strategy is that his "improvement" wasn't something that he maintained, once he reached his goal of a big payday, and the status of Expert. If he got a job, and then maintained that level while working, then that would have been impressive, but his main goal was winning a World Open. MDLM pulled out of some previous tournaments after a couple of early bad rounds; his goal was much more of a tournament result than a ratings goal. Somehow people looked past what his goals were. I've been 2010, he hit 2040. I guess MDLM is getting credit for achieving this goal faster (?), but everyone on this blog has already spent forever on chess, in practical terms. MDLM perfected his tactical system, but it's not as though he made Master; he might have made Master, but for GM he would really need that extra appreciation for all facets of the game.
@ LinuxGuy: Excellent observations - thank you!
ReplyDeleteAnother example (Problem #76) from Impove Your Chess Tactics - 700 Practical Lessons & Exercises - Yakov Neishtadt, New In Chess, © 2011.
FEN: 4r2k/p1p2rpp/1p3n2/6R1/2P2P2/PP1Q1P1q/1B5P/6RK w - - 0 1
Master Neishtadt provides the following instruction:
Assess the continuation 1. Rxg7 Rxg7 2. Bxf6.
Is the WBf6 absolute pin on BRg7 (BRg7 is also attacked by WRg1, so g7 is B.A.D. [1:2]) sufficient for White to regain the sacrificed Rook and win?
The B.A.D. ratio should have been [2:1] {in Whote's favor} in the previous comment:
ReplyDelete". . . so g7 is B.A.D. [2:1]"
Sorry about that.
GM Nimzovitch (in My System) wrote a nice “story” (System 1 loves to fabricate stories that make sense within any given context!) about the Pawn’s “lust to expand”. This description encompasses several features in a single image. The advancing Pawn can sweep enemy pieces out of the way, it can sacrifice itself to open lines of attack and it gets closer to promotion. Another way to look at it is that the potential gain in material (from Pawn to Queen) exerts a “gravitational pull” forward toward the 8th rank. This is why attention must be given to the possibility of Pawn promotion when any Pawn that has reached (or can be “seen” to reach) the 5th rank or beyond.
ReplyDeleteIn the same way, the formal rules of chess (i.e., the end goal is to attack the enemy King in such a way that its capture cannot be prevented) create another “black hole” (hidden “gravitational pull”) in the vicinity of the King, regardless of where it is located on the board. Those formal rules also create some anomalies. For example, because the King cannot be moved into check, there is a potential impact on sequences of exchanges. If there is a capturing sequence that requires the King to capture but the King would then be in check, that sequence cannot be done. We KNOW this rule, but we rarely recognize (think about?) the distortion it can cause while trying to calculate what to play.
In Problem #76, this rule is an integral (hidden) aspect of the solution. Master Neishtadt gives a plausible sequence of moves to begin the “thinking process.” This is a red herring, but a very potent one, because it primes our thinking to try to “confirm” that there is a subsequent sequence of moves that will work for White. After all, White is to move initially, and through lots of reinforcement, we are predisposed to think that whoever has the move in a given tactical problem MUST have a favorable solution, no matter how deeply hidden. The only way we can successfully apply Popper’s falsification is if we can overcome this natural tendency to look for confirming evidence.
The BRg7 (after the initial moves) is absolutely pinned to the Black King. It is absolutely immobile. However, by the formal rules, it can still prevent the enemy King from moving to any square within its attacking range. As a result of the initial moves, the g2 square has become B.A.D. The Black Queen (prior to this move sequence) had no possibility of doing anything useful against the White King. Now, it CAN do something useful: it can FORCE the WRg1 to move off the back rank with 2. …. BQg2+. Our minds (System 2) thinks there is no problem because the WRg1 can capture 3. WRxg2. After all, the BRg7 is still absolutely pinned. BUT, now comes a back rank check 3. … BRe1+. Again, the logical conclusion is that this can be covered with 4. WRg1, blocking the check – and the BRg7 is still pinned. The “surprise” occurs with 4. … BRe1xg1#. (I intentionally did not address the possibility of throwing a momentary roadblock in the way with 4. WQf1; Black simply captures it with check 4. … BRxf1+, and the final sequence is merely delayed by 1 move.) Note that White could not move his King at any time during this sequence because of the BRg7 – absolutely pinned but still very potent in restricting the White King’s options.
Delving back into Dr. Kahneman's Thinking, Fast and Slow, I started paying more attention to his references to the work of Dr. Gary Klein regarding intuition. Dr. Klein and Dr. Kahneman did some collaboration over about 7-8 years, resulting in an interesting joint article titled: Conditions for Intuitive Expertise: A Failure to Disagree. The article can be found here:
ReplyDeleteConditions for Intuitive Expertise: A Failure to Disagree
Dr. Klein has also written some apparently very good books on intuition and how to improve it. As I get time and spare money, I'll try to investigate his books.
I'm going to read it tonight.
DeleteI did do a few blitz tactics yesterday, in order to get a sense of where I'm standing. I should do a few hundreds more to have a more exact indication, as Aox pointed out. But as I already suspected, there is a transfer problem. So we solved only half of the riddle.
ReplyDeleteI have been able to convert knowledge into intuition. Now we have to find out what the right knowledge is to support transfer of intuition from one problem to another. The default base of 500 problems begin to become well digested. Now I'm going to compare them with the last failed problems. In order to find out which knowledge is needed and why it is not in my default base set of 500 problems.