PoPLoAFun REDUX
In search for salient cues, I stumble again on my PoPLoAFun system.
r1b2rk1/1pq1bppp/p3pn2/3n2N1/3R4/2N1B1P1/PPQ1PPBP/R5K1 w - - 0 15
Then have a look at the lines of attack.
And the Functions (defenders, overloaded pieces etc.)
And here you have it. Way too much information. And somewhere in this mess there a few essential salient cues. Less is more. When we can find a methodology to restrict the mind to the essential information, we might come further.
PoP = point of pressure
LoA = line of attack
Fun = function
Take the following position.
 |
| Diagram 1. White to move |
r1b2rk1/1pq1bppp/p3pn2/3n2N1/3R4/2N1B1P1/PPQ1PPBP/R5K1 w - - 0 15
Ni H. vs. Adly, A., Shenzen 2011
Let us first have a look on the balance of the points of pressure.
 |
| Diagram 2. Same position. PoPs. White to move |
Then have a look at the lines of attack.
 |
| Diagram 3. Same position. LoAs. White to move |
And the Functions (defenders, overloaded pieces etc.)
 |
| Diagram 4. Same position. Functions. White to move |
And here you have it. Way too much information. And somewhere in this mess there a few essential salient cues. Less is more. When we can find a methodology to restrict the mind to the essential information, we might come further.
Where to start? If there are any forcible imbalances in the position, then I suppose that the overloaded defenders should lead us the way.
The most obvious starting point (for ME) is the d5-square. Attackers/defenders on the d5-square are [3:2] before considering functions. The BNd5 "protects" BQc7 and BNf6, thus it already has two functions. The BNf6 “protects” BNd5 and h7, thus it already has two functions. The WNc3 appears to be pinned but it also can discover a double attack on the BQc7 if it captures on d5, with a Zwischenzug capturing the BBe7 with check if the WQc2 is captured. Given that the BRa8 and BBc8 are totally out of play regarding the kingside, White has local temporary superiority in the center and on the kingside. Attack the weakest point where you have superiority!
ReplyDeleteYou already drew the lines of attack. After 1. Nxd5 Qxc2 2. Nxe7+ Kh8, the Black king is FORCED into the corner (with NO MOVES!) AND 2. Nxe7+ also clears the LoA to the back rank for the WRd4. With the Black king immobile in the corner, it is easy to “SEE” that the BRf8 can be diverted off the back rank with 3. Nxf7+ Rxf7 (forced). Mopping up occurs after 4. Rd8+ Rf8 5. Rxf8+ Ng8 6. Rxg8#. Black resigned after 15. Nxd5!
That sounds too simplistic; I must be missing something vital.
The point is to “SEE” the critical “chunk” as the starting point of consideration; the weakest square of the position will almost always be part of that starting point. It is not the B.A.D. h7-square [2:2] because the LoA of the WQc2 can be closed easily, eliminating the potential Q+N mate on h7 (even presuming the BNf6 can be destroyed or diverted with tempo).
I have the feeling I'm trying to work miracles with System 2.
ReplyDeleteI suggest revisiting your post: How to feed system 1,
DeleteOctober 08, 2020.
A few thoughts about what I read on the internet.
ReplyDelete* the older world top (Kasparov cs) complains that the new generation have a lack of chess knowledge. They are only better at calculation due to working with the computer
* Carlson says that he is bad in chess puzzles
* Sindarov trains without a computer with a physical board
My conclusions:
Knowledge is not important for the trick.
Context is paramount for narrative building.
A puzzle provides too little context.
Knowledge books provide too many context.
Temposchlucker wrote (in the post and in a comment):
ReplyDeleteAnd here you have it. Way too much information. And somewhere in this mess there a few essential salient cues. Less is more.
I have the feeling I'm trying to work miracles with System 2.
I’m not trying to chide you, BUT...
You have “way too much information” because you are using System 2 to mechanically mark ALL PoPs, ALL LoAs, ALL Functions—before you allow yourself to experience an “intuition” (based on pattern recognition) from System 1, pointing to a particular local area that is “interesting” for some unknown (to System 2) reason as to what the position is really about. You ASSUME that you must thoroughly identify ALL these things in order to get System 1 to work its “magic.” That is a WRONG assumption.
The combined system of the mind [System 1 PLUS System 2] doesn’t work effectively like that. All that is needed to start the associative memory (System 1) working is recognition of a single point of interest. Think of that single point of reference as being like a rock [the single recognized pattern] thrown into a body of water [the “pool” of patterns that we have accumulated over time]. As soon as the rock hits the water, ripples are spread from that point of contact in all directions. This corresponds to recognition of adjacent patterns being activated but (perhaps) at a lower intensity level, just as the ripples start out large and then diminish as they move away from the initial contact.
I suggest you STOP trying to be so thorough. This is NOT an academic exam; there is no need to be thorough. Relax and allow your mind [System 1] to work its “magic” without insisting on System 2 controlling everything. If you will, DAYDREAM about the position without having any specific goal in mind. Our usual adult approach is: “I [System 2] MUST SEE and catalog everything using a sequential logical process in order to make sure I don’t miss anything.” No, you don’t. What you need is an approach that is called BEGINNER’S MIND in Zen.
As an interesting aside, most Western adults misconstrue that concept because their assumed definition of “beginner” is “novice,” someone who is unskilled. On the contrary, Zen BEGINNER’S MIND requires experience and skill in switching from reliance on System 2 to allowing System 1 to “speak” to and for us (knowing that System 1 does not use language to communicate; that is a System 2 function because communicating using language is a sequential process). The purpose of the Zen koan is to cause the logical mind to “run into a brick wall,” metaphorically speaking, so that System 1 can be heard clearly.
I’m not touting learning Zen or any other “religious” idea. (You cannot learn Zen using System 2 anyway.) I’m merely suggesting avoiding using System 2 to enforce control over the process.
Relax and enjoy the process unfolding on its own. I guarantee that if you just sit quietly and contemplate the position “as it is,” your System 1 will begin suggesting a point of interest (something that you subconsciously recognize) and then it will suggest other connections spreading out from that point of interest. And, if you allow it, you will find yourself in a state of “flow” with time passing without any consciousness of it. On the other hand, if you won’t allow it, it will prove to be very frustrating!
I understand your confusion.
ReplyDeleteYou are quite right when we talk about the tournament hall. No system will help you during an OTB game. Since there is no time for System 2 fiddle-faddling around. I never used PoPLoAFun or CCT during an OTB game.
In the study room anything goes. Since we need to discover things.
The situation is totally different from 2020. Back then, I had no idea how to educate System 1. Nowadays, that is totally clear, and I train that way on a daily basis.
You can't investigate without excessive use of System 2.
The remark about working miracles by System 2 was with tongue in cheek.
What am I looking for? I abandoned trial and error because it isn't productive in a bushy tree of analysis. You get lost in the thickets. I abandoned CCT because it has too much redundancy. I abandoned the use of intuition as the main guide in the study room, because that only works AFTER you have educated your System 1. We are in the position before that.
So I'm looking for a system that is able to prune the tree of analysis in a more efficient way than CCT.
I'm sorry: I did not "SEE" your tongue in your cheek. ;-)
DeleteText is a medium with a small bandwidth. And by explaining it with an emoticon it ceases to be a joke, usually. So I use them sparsely.
DeletePART I:
DeleteThinking last night, I came to a realization as to WHY CCT (Check, Capture, Threat) or the more robust CMCT (Check, Mate threat, Capture, Threat) does not work as a guide to the salient features of a position, and consequently does not “prune” the potential tree of variations.
[I’ll use the CMCT acronym throughout the rest of this comment.]
CMCT is a hierarchy of the degree of FORCING associated with each type of moves. The check is the most forcing move and MUST be immediately answered as part of the formal rules of chess. The other three levels of forcing moves do NOT require an immediate answer. This is how we get the heuristic regarding intermediate moves (Zwischenzug): the notion of the Equal or Stronger Threat, which is in order mating threats, gain of material and other threats.
Chess literature records that patzers seem follow the heuristic “Always check: it might be mate!”, taking “check” to mean a literal check in the chess sense—a direct attack on the enemy king. [That assumes, of course, that the patzer never realizes that he should “check” to “SEE” if there is a potential mating pattern lurking below the surface.] So, “Patzer sees a check, patzer gives a check!” in the chess sense.
Since that is the most forcing move in chess (the player with his king in check MUST do something to negate or eliminate the check; inability or failure to counter a check means the game is over: checkmate), it seems “obvious” that the first thing to look for in a position are all possible checks—“just in case it’s mate!” However, this apparent reduction in potential moves has no meaning beyond the immediate contact between the checking piece and the enemy king. It takes into consideration none of the other features of the position, much less the salient features that will strongly reduce the size of the forest of variations. As we learn from bitter experience, chasing the enemy king with checks with no other purpose is usually a waste of time.
The next lower level of forcing is the mating threat. It takes priority over gain of material or other kinds of threats: only an immediate check can counter a mating threat. Just as looking for immediate checks is counter-productive, so is looking for immediate mating threats without any other considerations.
The next lower level of forcing is gain of material (captures). IFF there are no checks or mating threats to counter, then a player can drop down to the level of gaining material.
Finally, the lowest level of forcing is “ordinary” threats. This level has the largest number of elements—everything else that is not included in the higher levels of forcing. This includes direct tactical themes/devices (fork, skewer, etc.) as well as deferred threats.
It is important to note that the intensity of forcing is a very poor selection guide for moves. It provides no guide for pattern recognition based on the interrelationships between the pieces and between pieces and squares.
PART II:
DeleteUsing CMCT is a demonstration of the substitution bias (Kahneman). We want to find the best move sequence. System 1 can’t find an easy obvious response and so substitutes a simpler question: “What’s the easiest approach to finding the next move?” The System 1 answer is “Trial and error, improved by looking for CMCT.” Unfortunately, System 1 does NOT tell System 2 that a substitution has occurred and it answers with great confidence. Our lazy System 2 nods approval (“Sounds like a plan!”) and goes along with the suggestion. We are then puzzled when it does not produce good results.
I had a copy of Charles Hertan’s book Forcing Chess Moves: The Key to Better Calculation. Here’s the advertising blurb on Amazon:
“Every chess fan marvels at the wonderful combinations with which famous masters win games. How do they find those fantastic moves? Do they have a special vision? And why do computers outwit us tactically? This rich book on chess tactics proposes a revolutionary method for finding winning moves. Charles Hertan has made an astonishing discovery: the failure to consider key moves is often due to human bias. Your brain tends to disregard many winning moves because they are counter-intuitive or look unnatural. We can no longer deny it, computers outdo us humans when it comes to tactical vision and brute force calculation. So why not learn from them? Charles Hertan's radically different approach is: use computer eyes and always look for the most forcing move first. [In essence he’s stressing using CMCT to guide the search.] By studying forcing sequences according to Hertan's method you will develop analytical precision, improve your tactical vision, overcome human bias and staleness, enjoy the calculation of difficult positions.”
Here’s WHY we should NOT try to slavishly ape the computer process: we don’t have the capability to a similar process at all!
Using forcing moves without any recognition of patterns of weakness and patterns of relationships that take advantage of those weaknesses is only available to a computer that can use trial and error (with a modicum of guidance in the form of heuristics) blindly searching every possible move deep into the tree of variations until the best end score bubbles back up to the ply 0 level. Given the millions of positions that must be visited and evaluated at a depth of search far beyond human capacity, that brute force search is a foolish approach for a human being. A trial and error approach based on forcing moves is only slightly better than just basic random selection of moves using trial and error—for humans.
Temposchlucker observed:
So I'm looking for a system that is able to prune the tree of analysis in a more efficient way than CCT.
I have some thoughts on a sound basis for a more efficient system than CMCT (forcing moves). I’ll defer that observation for now because of personal time restriction; I have lots of things on my “Honey Do” list for today.
FEN: r3k2r/pppn1pbp/1nb1pqp1/8/3P4/1QN1BN2/PP2PPPP/3RKB1R w Kkq - 1 12
DeleteI posted about this position during early May, 2024. The game score is at the end of the comment. I think it’s an excellent example of my hypothesis for pruning the size of the tree of variations.
How do we know what the salient cues are in a given position? IMHO, those cues are patterns of interrelationships of pieces and squares that signal some type of weakness, vulnerability, lack of coordination/cooperation, insufficient mobility, in short, anything and everything that can cause problems. Presuming prior training exposure to those types of cues, System 1 SHOULD signal recognition of the underlying pattern AND an associated way to take advantage of it.
In the example, there are two distinct areas (chunks?) that SHOULD trigger attention: the complex around the d5-square and the vis-a-vis of the BQf6 +BBg7.
The first pattern revolves around the d5-square. At present, it is not B.A.D. [2:3] in the formal sense of counting all IMMEDIATE attackers/defenders with legal moves to d5. However, in the sense of “looking through” obstacles, we can “SEE” that the d5-square is functionally B.A.D. [3:3] AND that it is “save” to advance the d4-pawn.
The second pattern involves the BBc6: it has no SAFE moves other than capturing on f3! We can “SEE” immediately that advancing the d4-pawn accomplishes two things: it “attacks” the BBc6 (having learned that “immobile pieces MUST be attacked”) AND it removes the possibility of capturing on the f3-square.
The third pattern is a reduction in mobility of the BQf6: it can only move to f5 or back along the h4-d8 diagonal. We can “SEE” that the BBg7 “protects” the BQf6 but that will be cold consolation if the Black queen is trapped by a minor piece with no retreat. What is more difficult to “SEE” is that the BQf6 “protects” the BBg7 and is very limited in its ability to do so: the f6-square is the only available square for that function.
It is not important to become aware of all those considerations at once, nor is it important for any particular order of becoming aware of all those considerations.
Those interactions are what determines the plan to be used to take advantage of as many of those factors as possible. In this case, the end result is gain in material. It is this process (not a formal step-by-step checklist process) that we have to learn to gain SKILL.
GM Rowson noted that GM Korchnoi “frowned” when he (Rowson) made his 11th move, as if to say “That’s not right!” He took a minute or two to put all those factors together and then confidently played 12. d5!
[White "Viktor Korchnoi"]
[Black "Jonathan Rowson"]
[WhiteElo "2568"]
[BlackElo "2558"]
[Date "2004.9.29"]
[Round "5"]
[Result "1-0"]
[Event "Monarch Assurance"]
[Site "Port Erin IOM"]
[Eco "A15"]
1.Nf3 Nf6 2.c4 g6 3.Nc3 d5 4.cxd5 Nxd5 5.Qa4+ Bd7 6.Qb3 Nb6 7.Ng5 e6 8.d4 Bc6 9.Be3 Bg7 10.Rd1 N8d7 11.Nf3 Qf6 12.d5 Nxd5 13.Nxd5 Bxd5 14.Rxd5 exd5 15.Bd4 Qf5 16.Bxg7 Rg8 17.Bc3 Qb1+ 18.Qd1 Qxa2 19.e3 O-O-O 20.Be2 Nc5 21.O-O Ne4 22.Qc2 Rd6 23.Bd4 Re8 24.Ne5 f6 25.Nd3 Rc6 26.Qd1 Qa5 27.b4 Qa2 28.Nc5 a6 29.Nxe4 dxe4 30.b5 Rc2 31.bxa6 c5 32.axb7+ Kb8 33.Bxf6 Rd2 34.Bc4 Rxd1 35.Bxa2 Rxf1+ 36.Kxf1 Kxb7 37.Bf7 Rf8 38.Bd5+ Kb6 39.Bc3 Kb5 40.Bxe4 Kc4 41.Be5 Re8 42.f4 Kb3 43.Ke2 Rd8 44.g4 c4 45.f5 gxf5 46.gxf5 c3 47.Bd4 Rc8 48.Bd5+ Kb4 49.Kd1 1-0
An interesting aphorism: "Pattern recognition is prediction."
DeleteAristotle — “We are what we repeatedly do. Excellence, then, is not an act, but a habit.”
DeleteMastery in any area comes not from a single moment of brilliance, but from repeated, consistent practice. When you practice a skill regularly, your brain rewires itself to perform it more efficiently. Over time, the actions become automatic, freeing up mental resources for higher-level thinking.
Small, consistent actions compound into significant results.
Good habits (like correct technique or disciplined routines) are easier to maintain than bad ones.
Practice makes perfect — repetition refines performance until it becomes second nature.
How to Apply This
Start small — pick one small behavior or skill to focus on.
Sequence it right — place the skill in a routine so it becomes automatic SUCCESS.
Be consistent — repetition is key; even short daily practice builds habits.
Track progress — notice how the skill becomes easier and more natural over time.
Thus spake the Sage Copilot.
Confucius:
I HEAR and I forget.
I SEE and I remember.
I DO and I understand.
JUST DO IT — Understand the trick!
The salient point is that Qf6 -Bg7 is vulnerable for a Röntgen attack by Bd4. The rest you can find by logical reasoning.
Delete* the clearance of d4
* the series of trades that prevent e6 from blockading the LoA by going to e5.
I found that this is true for most combinations. When you SEE the resulting tactic, you can find the path by logical reasoning. The tactic that makes the whole combination visible isn't necessarily quite at the end. I will give an example of that in the comments of a previous post.
I look forward to that example!
DeleteApparently you didn't see it. It was about diagram in the post about the FEN viewer:
DeleteWhat I need to see is the salient knight fork on f6 against Kg8 and Qd7. From there you can find the rest by logical reasoning.
When I cannot find this salient tactic, I will not find the tactic during a game when this position should arise.
PART I:
DeleteI do try to keep up with your comments in previous posts. If I recognize a different comment count, I go back through the posts to see what you (or someone else) has added. Somehow I missed that one added to the FEN viewer post.
Temposchlucker said:
What I need to see is the salient knight fork on f6 against Kg8 and Qd7. From there you can find the rest by logical reasoning.
When I cannot find this salient tactic, I will not find the tactic during a game when this position should arise.
IMHO, the salient knight fork tactic cannot be easily seen because one of the prongs of the fork is a surface-level clue (WNe4-f6+) whereas the second prong of the fork (WNf6xBQd7) is nowhere near the surface; it occurs 5 ply in the future AND BQe7 is NOT forced to recapture on d7 setting up the fork; BQe7 can escape to h4 allowing a loss in material but foiling checkmate with retention of the queen for counter-attacking. The displacement in time obscures the tactic.
Additionally, the knight fork on f6 is not the end goal; it is a means to an end. The end goal is a potential checkmate with the bishop and rook. In order to create that checkmate pattern, the g7-pawn MUST be “moved” to f6 without loss of momentum, opening the line of attack for the WRh1-g1-g8 and the WBc3-f6.
PART II:
DeletePerhaps that is the pointer to what must be seen: What rearrangement of the pieces will create/uncover a second weakness? If one prong of a double attack is plainly visible on the surface, that is a weakness even if there is no way to directly attack it in the present position. Every weakness should trigger a search for a potential second weakness AND ways to attack that weakness. The weakness of the c3-h8 diagonal is a surface-level clue. If the g7-pawn were already on f6, that would NOT allow checkmate by WRh1-g1+; BKg8-h8 gets out of check and WBc3xf6+ allows the bishop to be captured 3 different ways. WNe4 and WBc3 are the only attackers on f6 in the position as-is. Recognition of the potential checkmate pattern (WBf6+WRg1; BKh8) means that the White knight is the only possible sacrificial lamb to throw on the f6 altar to open the line of attack for the White rook.
Attention should shift to removing the defenders because there are no more White direct attackers that can be added to the “attack” on f6. Forcing captures are available on c7 (WQb7xBRc7 diverting BNd5 from defending f6) and d7 (WRd1xBNd7, attacking the BQe7 and [possibly] diverting it from defending f6). At this intermediate position, there is a branching choice: Black must decide to either give up material or be checkmated with a stock mating pattern. His choice is NOT forced.
ASIDE: When I refer to a double attack, I have in mind GM Averbakh’s definition (including threats of various orders) rather than the classical meaning.
In a sense, this is playing positionally rather than combinationally. The contours of the “final” position/pattern/solution are seen (dimly, as through a dark glass), and then System 2 logic works backwards to find a path/plan that will connect that visualized position to the current position—in reverse, recognizing the effect(s) of other local patterns. Along that discovery path, System 1 provides pattern recognition which supports the initial impression because System 2 has directed and focused the entire mind’s attention on solving a particular problem. Not initially focused on solving the overall tactics problem, but instead focusing the search for more salient features (weaknesses of all types and the means to exploit those weaknesses) that are hiding below the surface level (especially, hidden until a future point in the cogitation) AND which are connected (or can be connected) to the envisioned final pattern.
ASIDE: One of the “problems” with verbal descriptions is the sequential nature of them. Even if unintended, there is a tendency to see (in a verbal description) a sequence of patterns/actions (THIS recognition happens, then this recognition happens, etc.) where the patterns/actions may have been triggered simultaneously. I know no way around that verbal limitation.
We must discover the salient tactic IN THE STUDY ROOM. We are allowed and even encouraged to use al means and methods that are at our disposal. We shouldn't worry how System 1 works its miracles with the result.
DeleteIn the comment on diagram 1 from the FEN viewer, I described what prevented me from finding the solution. As you pointed out, that should not be our starting point. The starting point should be the mate indeed.
DeleteBack in 2020, my study tempo was effectively about 1 problem per day at average.. Since about a month or two, the average tempo is more like 7 problems per day. I have become much more effective already. Understand the trick! My posts explain it all.
ReplyDeleteIt is all about understanding. It is easier when you know what you are looking for. Just in case you missed the pun :P
DeleteCCT in itself is brilliant of course. Since every "technique" needs an additional tempo. But often it has too much redundancy. It gives false flags. If I could combine it in some way with the PoPLoAFun system, it might become more useful.
ReplyDeleteIn the end, it must translate to a skill. Hence the joke about working miracles.
DeleteUsually, there is one central idea in a combination. The idea that mates or gaines wood. The rest is designed to make the central idea possible or to clean up the mess afterwards.
ReplyDelete