Ceteris paribus
Robert stated:
"If the patterns (principles/”rules”/whatever) are NOT embedded in LTM, then it is highly likely that we will NOT “SEE” the available options."
This is the core conclusion of this blog. It might not be immediately obvious what this means for our training. The following diagram makes this perfectly clear:
 |
| Diagram 1. Black to move |
r3r1k1/1p3pp1/p5N1/P1p3n1/6b1/2N1P1B1/1PQ2PKP/R4R2 b - - 0 22
Pal Benko vs. Wolfram Hartmann (1984)
I hesitated between Bh3+ and Bf3+. I thought it was equal, ceteris paribus. Admittedly, it was 6:00 am this morning and I just woke up. But that are the perfect moments to show you what has been absorbed and what not! What a grandmaster has absorbed doesn't disappear when he is drunk or sleep-drowsy.
As long as these two moves look the same at first glance, I haven't absorbed the patterns. And that has dire consequences! What you haven't absorbed, you can't SEE during calculation.
Diagram 2 is from my game yesterday. I tried to calculate the bishop sacrifice Bxh7+, but I couldn't work it out until the end, so I played Rg3 instead. That is still winning of course, but that is not the point. The point is you can't SEE what you haven't absorbed. You cannot calculate what you can't SEE. And we talk about very simple things here, as shown in diagram 1.
 |
| Diagram 2. White to move |
r1b2rk1/ppqn1ppp/1n2p3/2ppP3/5P2/1P1BPR2/PBPN2PP/R2Q2K1 w - - 1 13
It took me another 14 moves to win the game.
The message is, nothing is too trivial. Use precise observation as your guide to hunt down omissions in your absorption. Don't feel embarrassed, just fix it.
ReplyDeleteStrange: one of the salient "cues" for the success or failure of the Greek Gift in this specific position is the BQc7: LPDO.
ReplyDeleteIn an earlier post (2016), you opined:
"Is it really possible that grandmasters calculate everything until quiescence? I doubt it. They calculate to the next steppingstone.”
It's fairly easy to mentally sacrifice the Bishop Bxh7+, followed by Rh3+. It's also fairly easy to evaluate that resulting position as favorable for White: White has two major pieces bearing down on the (relatively) unprotected Black King, and there are four "reserves" waiting to join in—WPe5, WBb2, WNd2, WRa1. That e5-pawn is a key component of a successful Greek Gift. Most importantly, White has the initiative. Black’s army is over in the queenside “wilderness” doing nothing to assist their King, so White has local superiority on the kingside. Strike while the iron is hot!
GM Tisdall (analyzing a different position) said that you should go ahead and get that steppingstone position actually on the board because you have reasonable certainty that it is favorable. But that’s the scary part: we aren’t 100% sure that it’s favorable because we don’t have the followup pattern(s) embedded in LTM that scream “JUST DO IT!”
My initial reaction was like yours: I was not sure what to do after only two moves. Then I decided that GM Beim’s advice applied in this steppingstone position: "If it doesn't work, but you really want it to, then it must work!"
It is a little more work to get to the next steppingstone position, but not too hard to “SEE.” The White Queen will get to h7 with check. Black MUST be able to escape from the check. The only viable way of doing that is to open the escape hatch by moving the BPf7. That brings into play the WPe5 (capturing) backed by the WBb2. More lines are ripped open for the two White heavy pieces.
“De l’audace, encore de l’audace, et toujours de l’audace” (“From audacity, more audacity, and always audacity”. Attributed to Georges Danton during the French Revolution 1793. (You can substitute “initiative” for audacity.)
The final nail in the coffin for Black is that unprotected Queen on c7. If the Black King is on his second rank and the WRg3 can capture on g7 backed up by WQh7, Black is going to lose the Queen. That’s what I had a hard time “SEEing” at the initial steppingstone position.
If audacity means just gambling, then it is just plain silly. Often it takes courage to not gamble. My logic is: when it is too complicated, improve your position. And that is what I did.
DeleteI looked at f5 by black, closing the diagonal. I somewhere learned the rule to "always" take exf5e.p.. But Stockfish doesn't agree. Just adding th knight to the equation is enough.
Moves in such positions tend to be double edged. It is about the balance between opening LoAs for your attackers, and giving your opponents defenders more space. Sometimes NOT opening LoAs and keeping your opponents defenders clogged up is better. There are no universal rules.
So in the end it is about absorbing simple patterns (principles/”rules”/whatever) in the study room.
“(You can substitute “initiative” for audacity.)”
DeleteI don’t see that as gambling, but I did counsel “gambling” when it comes to testing my opponent’s skill in being able to force checkmate in 50 moves in the K+B+N vs K endgame, so I guess that's a reasonable inference to make. I am usually risk-averse while playing.
Usually it is a cyclic process. When I take too much risk, I end up playing a few games while taking too little risk. Until I'm fed up and start to take more risk again.
DeleteI haven't developed a style actually. Logic guides my experiments. I played gambits for 7 years, just to get better at tactics, not because I especially liked gambits. I started with 1 year, which is the general advice, but since I hadn't the feeling I learned enough from it, I ended up with adding another 6 years.
Now I play more solid, which is more in accordance to my nature, but my solid is not the same as other people's solid, since I obey no rules, only logic.
And logic is destructive by its very nature, you can only disprove thinks. Hence I work by elimination.
I don't lack the courage to follow my logic, as far as I know. If that is audacity, then I'm good, I suppose.
The most important thing is that I learn something from what I do.
But we are talking about the wrong diagram here. Diagram 1 is the one that points to the problem.
ReplyDeleteI saw a picture with only the upper two centimeters of a person's hair, and that was enough to recognize the person. That shows how incredibly capable the fusiform face area (FFA) is in recognizing patterns. The latest discovery is that the FFA is used for the recognition of all kinds of patterns (principles/”rules”/whatever).
Bh3+ and Bf3+ are NOT the same, and the FFA has the capability to recognize the difference immediately, but it didn't. My training using this problem has not ended until it does so.
On Chess Tempo, I scanned the Tactical Motifs and Themes tags and did not find the specific mating situation given above in Diagram 1. Please correct me if I missed it.
ReplyDeleteI suggest explicitly noting the circumstances and mentally tying that context to the salient parts of the position. Three squares must be covered in order to checkmate: g2, h1 and g1. The White pawns and Rook occupy three of the squares surrounding the White King. The Knight can only attack one square of the killbox – g1. The Bishop MUST attack the other two squares. Since the Bishop covers only white squares, it must cover g1 and h1 simultaneously leaving the g1 square to be covered by the Knight. Ergo, the Bishop moves to the f3-h1 diagonal with check 1… Bf3+, followed by 2. Kg1 Nh3#. I suggest thinking of “Place the Bishop on the longest diagonal” as a “rule” to help trigger the appropriate response of System 1.
I know that sounds trivially obvious—and it is. However, it is vitally important to consciously focus attention on the contextual aspects when trying to cement a pattern into LTM. The combination of the visual pattern and the contextual “rule” should be sufficient for the FFA to do its magic. In order to cement that pattern, I suggest finding other examples if you can. Examples of unlabeled patterns are much more difficult to find. If you can’t find any, make up your own.
Errata:
DeleteSince the Bishop covers only white squares, it must cover g2 and h1 simultaneously leaving the g1 square to be covered by the Knight.
Sorry about that.
The problem is looking without SEEing. It is trivial indeed. That's why we tend to skip it. Thinking that we can see it at will when we set ourselves to it. And indeed, we do. But this is about SEEing without setting ourselves to it. Just as we can't prevent recognizing the person while seeing only the upper two centimeters of his hair.
DeleteThe book registered it as a suffocation mate. Which is different from a smothered mate.
DeleteBefore I forget:
DeleteWhen we “SEE” a particular sequence of scenarios that clearly leads to a win, it is pointless (except for training purposes) to point out that an alternative sequence MIGHT be shorter or more “beautiful”. The objective is to win; if that is accomplished, then “woulda, coulda, shoulda” is useless recrimination.
“A bird in the hand is worth two in the bush.”
The proverb advises caution against unnecessary risks. It suggests that a certain, smaller advantage is preferable to a speculative, larger one that may never materialize.
At any given point in time, we have a repertoire of patterns that we can use to project scenarios into the future. After the game, there is time for leisurely reflection on what might have been, with the intent to increase our usable set of patterns.
Onward.
I tried finding other examples of the pattern displayed in Diagram 1 on lichess and Chess Tempo, so far without luck. Lichess does not have the suffocation mate category. Chess Tempo has 181 problems categorized as suffocation mates. I’m working my way through them.
In a sense, it is a smothered mate where one wall of the killbox is formed by our bishop
DeleteThe point is, I see the bishop without noticing what it is actually DOING.
DeleteI found this example on Wikipedia:
DeleteSuffocation mate
The suffocation mate is a common method of checkmating. It works by using the knight to attack the enemy king and the bishop or queen to confine the king's escape routes.
FEN: 5rk1/5p1p/8/5N2/8/2B5/8/7K w - - 0 1
(I re-positioned the White Knight to the move immediately preceding the checkmate, and added the White King arbitrarily.)
I think this example triggers an affinity to the Pillsbury Mate because of the WBb2 on the long diagonal, cutting through the Black King position.
This Chessable course contains a chapter with suffocation mates.
DeleteThe link went to a 404 page. I presume that link was supposed to be to The Checkmate Patterns Manual. The Variation Details lists 10. Suffocation mate (16 variations).
Deleteindeed. Weird, I have no problems with the link.
DeleteFEN: r1b2r1k/1p2Bn1p/p1qP2p1/2pN4/3Q4/8/PP4PP/R4RK1 b - - 0 0
ReplyDeleteHere’s an example based on a similar motif to Diagram 1. The preceding move was 0… cxQd4. Note the importance of the long diagonal combined with the “hole” on g7.
FEN: 4rkr1/1p2qp2/p2p3Q/2pP2N1/2Pb1PP1/6R1/PP4K1/8 b - - 0 0
ReplyDeleteHere’s an example based on a similar motif in Diagram 2 at the end of the Greek Gift sacrifice. The preceding move was 0… Rg7. I infer that White previously played 0. Qh6+. The finishing “two-step” maneuver is something to remember as a pattern.
Sorry, not related to Diagram 2 but to this position:
DeleteFEN: 4r1rk/1p3p1p/p2p4/2pP4/2PbNPPq/6R1/PPQ3K1/7R b - - 0 0
Black played 0… Qe7.
1.Rxh7+ Kxh7 2.Ng5+ Kg7
[ 2...Kh6 3.Qh7# ]
[ 2...Kh8 3.Qh7# ]
3.Qh7+ Kf8
[3...Kf6 4.Qh6+ Rg6 5.Nh7#]
4.Qh6+ Rg7
[Notice the smother if 4...Bg7 5.Nh7#]
5.Qh8+ Rg8 6.Nh7#
Interesting maneuvers at each steppingstone position in the solution.
[Robert as Anonymous - Unable to sign in after logging out and back in multiple times]
ReplyDeletePART I:
Reference: https://www.researchgate.net/publication/28765443_The_emergence_of_choice_Decision-making_and_strategic_thinking_through_analogies
– Alexandre Lenhares—ResearchGate.net
According to Linhares (and chess experts), the following two problems are “strategically similar” pairs of positions:
Position 8:
FEN: 5r1k/5p1p/5P1N/8/8/8/4p3/K5R1 w - - 0 1
Position 20:
FEN: 1kr5/ppN5/8/8/5Qp1/1r1p2P1/2b2K2/8 w - - 0 1
Here we can observe that analogy plays a significant role in chess cognition—chess experts report that these pairs are ‘very similar’, despite the large number of superficial differences. The positions in each pair differ in piece sets, number of pieces, position of pieces, and the underlying search tree (from the point of view of the previous ply), so how can they be similar to each other?
Here are the definitions of suffocation mate (Position 8) and smothered mate (position 20) from Chess Tempo:
Suffocation Mate:
The suffocation mate is similar to a smothered mate where a knight delivers mate to the king on edge of the board, however instead of being completely hemmed in by its own pieces and pawns, the mated king is prevented from escaping by a combination of its own pieces, and the mating side's pieces cutting off any movement away from the edge of the board.
This tag should only be applied when the king is actually mated on the edge of the board. A partially hemmed in king that escapes away from the edge of the board, and is eventually mated elsewhere, or is mated by a piece other than the knight should NOT be tagged with this tag.
The main difference between a smothered mate and a suffocation mate is that in a smothered mate the king's own pieces are blocking all escape paths, while in a suffocation mate, some of the escape paths away from the edge of the board are blocked or controlled by the attacker's pieces.
Smothered Mate:
A smothered mate occurs when the king is trapped on the back rank or side file, and is unable to escape an attack from a knight due to being hemmed in by their own pieces. Escape away from the edge of the board is typically cut off by pawns or other friendly pieces, and escape along the edge of the board is often restricted by heavy pieces such as friendly rooks or queens.
This tag should only be applied when the king is actually mated on the edge of the board. A partially hemmed in king that escapes away from the edge of the board, and is eventually mated elsewhere, or is mated by a piece other than the knight should NOT be tagged with this tag.
The main difference between a smothered mate and a suffocation mate is that in a smothered mate the king's own pieces are blocking all escape paths, while in a suffocation mate, some of the escape paths away from the edge of the board are blocked or controlled by the attacker's pieces.
PART II:
ReplyDeleteLinhares asks: How can a computational model perceive such abstract strategic similarity?
I am relatively unconcerned with how computational models “perceive” but am very interested in how humans perceive and how humans can be TRAINED to perceive similarities and differences based on abstract analogies.
There IS a small difference in the tactical sequences of the two scenarios.
In position 8, there is a check on g8, forcing a capture on g8 and simultaneously removing the guard from f7, followed by a knight move to f7 giving mate.
In position 20, there is a double check (forcing the Black King to move), followed by a Queen sacrifice on b8 forcing a capture on b8 and simultaneously removing the guard from c7, followed by a knight move to c7 giving mate.
I have a simple question. Is this an example of a distinction without a real difference—for the purpose of training human perception of chess patterns?
For the purpose of training to “SEE” the cues and salient features based on patterns by analogy (comparing an actual position with a retained image/pattern), I don’t perceive any difference. The tactical action sequence seems to be almost identical. After an unspecified number of training examples, I think it highly likely that these two “differences” will blend together into a more abstract pattern for future recognition of the underlying pattern(s).
I perceive a difference between a piece that blocks its own king and an empty square that is blocking that very king. They are analogous, but not the same. You must have absorbed an extra action: what is your piece doing.
ReplyDeleteDiagram 1 shows this even more: I needed to imagine two extra actions:
*the aura of the bishop AFTER it moves to f3. This is not immediately perceived since the white king blocks the aura (from a perceiving point of view)
*the effect of movement of the white king
I must see what my bishop does AFTER it moves to f3 and AFTER the white king moves to g1.
As long as I see two different actions, I must calculate it (system 2).
Delete[This time, Blogger let me use my Google sign in; I did nothing different.]
DeleteNo argument; just a different personal perspective.
Does Dr. Kahneman's WYSIATI [What You See Is All There Is] apply?
Perhaps the word "conceive" should be used instead of "perceive" since the “recognition of pattern(s)” process is more of a mental conception rather than a physical perception.
I have a curious “recognition” response when I encounter patterns like Diagram 1, position 8 and position 20. I "SEE" more than just the starting move: I comprehend the pattern type from the perceived cues, and then (almost immediately) the action sequence of the move(s) in the beginning scenario of the expected principle variation become “visible.” If multiple scenarios in sequence are required to reach a “final” solution point, I have to clearly imagine the first steppingstone position, then “read” the cues in that imagined position, triggering the action sequence of moves in the next scenario; lather, rinse, repeat until an evaluation can be made which signals I have reached an endpoint.
I think that a lot of the recognition/response is very dependent on the level of abstraction that we “SEE.” I am no longer consciously aware of piece auras because I spent so much time practicing “looking through” the pieces to the edge of the board that it has become second nature.
When a particular pattern has been triggered, the next step is mental simulation using an action sequence. The action sequence in a given scenario is a few steps long, usually less than 6. As GM Aagaard pointed out, move-by-move calculation (System 2) is needed when we do NOT holistically “SEE” the patterns (including the associated moves). That lack of sight might be because the relevant cues and patterns are not ingrained, or it could be because of a “retained” image from the actual position to the steppingstone position or another in the sequence of scenarios, or any number of alternative explanations.
I think System 2 provides the transition from scenario to scenario, but that part of the “thinking process” is entirely opaque to me.
As I was told night before last in a public church meeting, my brain apparently does not work the same way as the brains of most people. My wife wholeheartedly concurs.
That has always been the problem with examples in this blog. Since an example is a personal observation, it is almost never recognized.
DeleteBut it is an invitation to observe at what trivial point you personally fail.
The good news is, that in all three games with white I played lately, I sacced the Bxh7+, and I won all three games. So apparently, the understanding of the patterns that are needed for this kind of attack is growing.
ReplyDeleteThe attack starts to become more LoA oriented. What is the danger of the black king accepting the sacced bishop on h7? That he kan skedaddle away via f7 or f8-e7.
DeleteNow I'm more aware of Nd2-f3-g5 covering f7 and a4-Ba3 covering e7 and f8.
While working my way quickly through Chess Tempo's Suffocation Mate problems, I've noticed that, quite often, the entire sequence of moves in the solution springs readily to mind; it is only recently that I've experienced that sensation.
ReplyDeleteYou are definitely making progress!
Soccer and chess training
ReplyDeleteThe grand scheme is:
ReplyDelete* whenever you have to calculate (system 2) there are things to be absorbed.
* start with the elements that are 1 ply deep, then 2 ply, then 3 ply.
* only go to a deeper ply level when you have ALL lower plies absorbed. This has to do with the frequency of occurrence.