What do the pieces do?
A close inspection of my last post revealed that I probably mix up two things. The plan of a tactic is revealed by its last scenario. The main study task is twofold:
- discover all scenarios in the chain of the logical narrative
- discover all the effects of each scenario
I mix this up with my observations of the games at the Tata Chess tournament.. Essentially that works the same, but the end of the scenario-chain is not the gain of wood, but an assessment of the resulting position. Which I called the plan.
Back to tactics. The last post was about the first point, discovering all scenarios, and especially the last scenario. But during my training I discovered that point two is a problem too, seeing all the ins and outs of a single scenario.
Let me give an example.
 |
| Black to move |
5rk1/p3ppbp/1q1pb1p1/1N3r2/2PB1Qn1/1P4P1/P3PPBP/R4RK1 b - - 0 1
Parligras, Mircea vs. Nielsen, Peter Heine
This is a problem of my database, and Chessable presents it to be on a spaced repetition basis.
I think I hadn't seen this position for two months or so.
What did I see immediately (what have I absorbed)?:
- 18. ... Bxd4 taking away the thread against my Queen
- 19.Qxd4 Rxb5 annihilating the defender of the white queen
- 19.Qxg4 Rxb5 discovered attack against the white queen
What did I not see (what have I not absorbed)?
- 18. ... Bxd4 comes with an additional punch against Ra1
- 18. ... Bxd4 four black pieces converge on f2
I this as a proof how to make progress in tactics. I must focus on one scenario at the time. Until I see what the moves actually do, how two scenarios interconnect to each other and how the pieces converge.
This is the picture now. The following holes in my bucket are more or less plugged:
ReplyDelete-opening white
-openings black
-positional play
Plugged, but not converted into a weapon yet.
The following hole is not plugged:
-endgames
My tactics are slowly being converted into a weapon. When positions become tactically complicated, I do usually better than opponents of the same rating. Which most of the time converts into a better ending (one or two pawns up, or positionally advantageous). Which I tend to screw up.
FEN:5rk1/p3ppbp/1q1pb1p1/1N3r2/2PB1Qn1/1P4P1/P3PPBP/R4RK1 b - - 6 18
ReplyDeleteI presume you’ve looked at the game score. White’s 18. Bd4?? was a move-order blunder; he should have played the alternative line given by GM Stockfish (in bold in the game score below), reaching a more or less equal position.
What is the significant difference between the two lines?
In the game continuation, White goes into a losing exchange sequence (losing WBd4+WNb5 for BBg7).
In the alternative continuation, White gets an even exchange (swapping WNb5 for BNg4). After 18. Qxg4 Rxb5, White can retreat with 19. Qf3, protecting the WBc3.
I have a hard time understanding why a 2600+ GM didn’t figure that out. It doesn't seem that complicated.
What am I missing?
Chess Tempo
[White "Mircea Parligras"]
[Black "Peter Heine Nielsen"]
[WhiteElo "2625"]
[BlackElo "2687"]
[Date "2011.9.5"]
[Round "3.3"]
[Result "0-1"]
[Event "FIDE World Cup 2011"]
[Site "Khanty-Mansiysk RUS"]
[Eco "A38"]
1.c4 c5 2.g3 g6 3.Bg2 Bg7 4.Nf3 Nc6 5.Nc3 Nf6 6.d4 cxd4 7.Nxd4 O-O 8.O-O Nxd4 9.Qxd4 d6 10.Qh4 Be6 11.Bg5 Qa5 12.Bd2 Rac8 13.b3 b5 14.Qg5 Rc5 15.Nxb5 Qb6 16.Qf4 Ng4 17.Bc3 Rf5 18.Bd4 ( 18.Qxg4 Rxf2 19.Bd4 Bxg4 20.Bxb6 Rxf1+ 21.Rxf1 axb6 22.Nc7 Rc8 23.Nd5 Kf8 24.Kf2 ) 18...Bxd4 19.Qxd4 Rxb5 20.Qd2 Rc5 21.Rad1 a5 22.Qf4 a4 23.Rd3 axb3 24.axb3 Nf6 25.h3 d5 26.cxd5 Nxd5 27.Qh4 Nf6 28.g4 h5 29.g5 Nh7 30.Rg3 Rb8 31.Rd1 Rd8 32.Ra1 Qc7 33.Bf3 Rc1+ 34.Rxc1 Qxc1+ 35.Kh2 Qc5 36.Kg2 Rd4 37.e4 Rd2 0-1
The game score isn't given in the Chessable course. What I hope you are not missing is the point I'm trying to make: the necessity to work at node level.
DeletePersonally, I find it reassuring that these types of blunders are still made at master level. Every day one or two blunders happen in the master + challenger groups at Tata.
I searched back through the blog for “node” so that I was sure I clearly understood what you mean by that term. I agree that the "working level" must be at the node level.
DeleteThere are three blog posts containing “node.” I picked the Lowest common denominator — February 26, 2024 post as most informative.
As I understand it, a node is a potential branching point, which changes analysis from a straight-line sequence of transformations (move-by-move) into a tree of transformations with two or more alternatives branching out from each node in the tree. That concept is applicable to both the Tree of Analysis and the Tree of Scenarios.
At a node, the alternatives are associated with more than just “pieces-on-squares” [POS]. The formal chess rules can come into play [i.e., that a King cannot move into check during an exchange sequence], Checks, Captures and Threats [CCT] (of all kinds including Averbakh’s broad definition of a double attack to include direct attacks and threats of various orders), prototypical tactical themes (pins, forks, skewers, etc.) and stratagems (such as the Greek Gift) can be applied, and so forth. In short, any and all patterns (KNOWN BY AND AVAILABLE TO THE PLAYER) can be used to guide the investigative process at each node, either requiring additional search or producing a cutoff of further investigation along that alternative.
The important thing at each node is to CHOOSE the most urgent variation (a “forcing” sequence is best for reducing cognitive load by pruning, but there are alternatives that are just as useful, such as “quiet moves”) and traverse it forward, using scenarios to “jump” from one node to another (perhaps separated by more than just the next move) until either (1) a definitive desirable result is reached (significant material gain or a known pattern from which win, lose or draw can be inferred) or (2) the variation abruptly terminates in a recognized refutation. This is primarily a “depth-first” search. As you noted, after reaching the end of the principal variation, then go back and determine what patches (if any) are required at any particular node(s). At this stage, it is vitally important to not by blinded by confirmation bias. The hard thing to do is to honestly apply Popper's falsification process.
During the investigation along the principal variation, “interesting” complications can be observed and resolved while at each node, rather than waiting until the end of the overall variation. Those “resulting moves” (GM Beim) must be accounted for in the principal variation, and may trigger a search for a better alternative at the beginning (or at least earlier in the principal variation), perhaps requiring a different principal variation. This is NOT a strict application of Lasker’s Rule: “When you see a good move, look for a better one.” If the complications increase without SEEing a resolution, that is a good indication that the principal variation may be fatally flawed.
Klein opines that usually there will only be about three (NOT A-L-W-A-Y-S!!) options to investigate at each beginning node, and each option will only have to be explored using mental simulation to a depth of about six (NOT A-L-W-A-Y-S!!) sequences of scenarios in order to find a viable option.
Have I missed anything?
I was referring to this post.It is insufficient when I don't see the additional punch to Ra1 when I play Bxd4. I can never hope to become good at tactics when I don't fix this.
DeleteIt are the simple things we are blind for. The good news is, that I don't need to learn complex things just yet. When I learn these simple things, it will already make a huge difference.
We must first learn to see the obvious.
ReplyDeleteSurfaces and essences – we have to start with what’s “obvious” as “clues” to be able to SEE what is not obvious.
ReplyDeleteI recently read an interesting idea.
The Prototype Principle—The most specific event can serve as a general example of a class of events.
“Everybody knows that specific events have a vividness which imprints them so strongly on the memory that they can later be used as models for other events which are like them in some way [similar in essence]. Thus in each specific event, there is the germ of a whole class of similar events. The idea that there is generality in the specific is of far-reaching importance.
D. R. Hofstadter, Gödel, Escher, Bach: An Eternal Golden Braid, pg 352
Add the frequency of occurrence to that and you have a tool for guidance.
DeleteIt is difficult to communicate this. Since the obvious that I don't see probably differs from the obvious you don't see. But the the message is, before you worry about the not so obvious, you must first learn to see the obvious.
ReplyDeleteTemposchlucker opined:
ReplyDelete"[B]efore you worry about the not so obvious, you must first learn to SEE the obvious." [Emphasis added.]
That sounds very similar to GM Aagaard's aphorism:
"Before 'thinking outside the box', learn to 'think inside the box'.”
In the given example, I was 'aware' of the two possibilities that didn't automatically spring into your consciousness. Given the principal variation that seemed ‘obvious’ (the gain of material on d4 and b5) after 18. Bd4, I just pruned them from further consideration. It was only after going through the game score and reaching the position prior to White’s 18. Bd4?? that I realized there was some significance to the BBg7 pin of WBc3 against the WRa1.
I do understand the difference between training and playing. For nearly as long as I have been studying and playing chess (over 55 years), I have tried to follow this maxim as closely as possible:
“Train like you play, and you will then play like you train.”
As you noted, what seems obvious to you may not be obvious to me, and vice versa. What I find 'curious' is that what seemed 'obvious' to me was NOT 'obvious' to a 2600+ GM. Surely there must be some contextual reason for his apparent 'blindness' leading up to the critical moment. Perhaps time pressure, or even simply a 'brain fart' illusion; “stuff happens” to the best of us.
We tend to overestimate the masters. The eval bar shows that they are blunderprone too. The reason for our admiration is based on not realizing how bad we are. Is this post shows.
DeleteSo we overestimate the masters because we overestimate ourselves. It even feels as heresy to write this down.
While cogitating on nodes, I realized something about the criteria for terminating further calculation that had previously escaped my attention.
ReplyDeleteQuiescence is the usual criteria for terminating further tactical search along a particular scenario or variation. Copilot opines:
A position is considered quiescent when the tactical storm has passed, meaning there are no forcing moves that could significantly alter the balance of the game. In such positions, static evaluation—considering material, pawn structure, piece activity, and king safety—becomes reliable. Conversely, a non-quiescent position is unstable, with ongoing tactical possibilities that make any immediate evaluation potentially misleading.
The criteria I have previously used to terminate forward search are (1) a terminal position is reached [win, lose, draw], (2) a static evaluation is made that a deeper search will not significantly change that evaluation, or (3) no further tactics are available for either player at this node (quiescence).
I overlooked that in a scenario-based search, it is possible to stop investigating deeper whenever the potential subsequent scenario(s) are known patterns (regardless of the required moves) with known outcomes. The following saying immediately came into conscious mind:
I know at sight what a position contains, What could happen? What is going to happen? You figure it out. I know it!
José Raúl Capablanca
It is similar to “knowing” the outcome of a potential Trébuchet position and how to play it out (regardless of the opponent’s moves), prior to the actual position arising on the board. The “knowledge” [skill?] extends all the way to the end of the game in spite of the combinative explosion of possible moves. Consequently, deeper analysis is not required.
In practice, there is often another criterium: this is too complicated for me, I just increase the pressure instead.
DeleteWhile studying on Chess Tempo, I was struck by Problem #80627350. There are two recognizable “patterns” that could trigger recognition and possible scenarios. It’s hard to determine which pattern would be triggered first. I suspect that the first scenario (based on a royal knight fork) might be the more obvious first choice to investigate. Why? Because of the thinking process heuristic “look at the opponent’s last move.” After checking that variation out (or possibly BEFORE checking it out), there may occur a realization that Black was down a Queen at the start of the puzzle. In the given puzzle position, there is a second scenario (not quite as obvious), but I think that makes the puzzle too easy). The second scenario would only be triggered if the Lasker heuristic “When you SEE a good move, look for a better move” came to mind.
ReplyDeleteWhen given only the position immediately preceding the combination, it is sometimes difficult (if not impossible) to figure out why that position arose. As you can see, I put the problem FEN last, preceded by the position 3 ply earlier. I’ve included both in the order they occurred in the game. I found the game score and included it below.
GAME FEN:2r2nk1/1Nr2ppp/p7/1p2N3/3q1n2/P5Q1/1PB1RP1P/R5K1 w - - 1 23
Problem #80627350 FEN: 2r2nk1/1Nr2ppp/p7/1p2N3/4qn2/P5Q1/1PB2P1P/R5K1 w - - 0 24
I think it illustrates why it is important to SEE the context preceding a puzzle. Otherwise, there is no realization of the cues that trigger the reaction. Once triggered, there is a typical scenario that flows from each variation.
[Event "ZMDI Open 2014"]
[Site "Dresden GER"]
[Round "8.9"]
[Date "2014.8.22"]
[White "Pirrot, Dieter"]
[Black "Teske, Henrik"]
[WhiteElo "2389"]
[BlackElo "2505"]
[ECO "C21"]
[Result "0-1"]
1.e4 e5 2.d4 d5 3.Nd2 Be7 4.Ngf3 Nf6 5.Bd3 c5 6.c3 cxd4 7.cxd4 dxe4 8.Nxe4 Nbd7 9.Nc3 a6 10.O-O b5 11.a3 Bb7 12.Re1 O-O 13.Bc2 Rc8 14.Qd3 Re8 15.Bg5 Nf8 16.Nxe5 Nd5 17.Bxe7 Rxe7 18.Ne4 Nf4 19.Qg3 Nxg2 20.Re2 Rec7 21.Nc5 Qxd4 22.Nxb7 Nf4 [GAME FEN] 23.Re4 Qxe4 24.Bxe4 [PROBLEM #80627350 FEN] Rc1+ 0-1
The problem with the problem FEN is that it is the wrong FEN. Since the computer moves first.
DeleteBut yeah, the knight fork comes first and then the back rank problems show itself.
I didn't catch that discrepancy until after I posted it. Chess Tempo gives the position after the ply immediately preceding the position that starts the problem. As soon as the problem is displayed, the computer advances 1 ply by playing the "bad" move that sets up the problem. Sorry about that.
DeleteHere’s an example of how someone fails to SEE a possible scenario while analyzing. Black is two pawns down, with no apparent checkmate or potential material gain to balance the position. The “idea” (combination of tactical motifs and themes) is not obvious on the surface. However, religiously considering forcing moves first might have given him a win instead of a loss. Let’s be generous, and assume that he was in severe time trouble.
ReplyDeleteWhat cue(s) did he miss?
IMHO, he did not remember the novice aphorism, “Always check; it might be mate,” especially when the enemy king is already immobile.
The result: White won a game he should have lost—in 5 moves!!
FEN: 3q4/2P3rk/4Q3/1r6/p7/8/P4R1P/2R4K b - - 0 44
[White "Anthony J Miles"]
[Black "Slobodan Martinovic"]
[WhiteElo "2500"]
[BlackElo "2460"]
[Date "1985.7.??"]
[Round "5"]
[Result "1-0"]
[Event "OHRA"]
[Site "Amsterdam NED"]
[Eco "E10"]
1.d4 Nf6 2.c4 e6 3.Nf3 a6 4.Nc3 c5 5.e4 cxd4 6.Nxd4 Bb4 7.Bd3 Qc7 8.O-O O-O 9.Bg5 Be7 10.Qe2 d6 11.Kh1 Nc6 12.Nxc6 bxc6 13.c5 d5 14.e5 Ne8 15.Bxe7 Qxe7 16.Na4 Ra7 17.Nb6 Nc7 18.Rac1 g6 19.Qg4 a5 20.Qa4 Bb7 21.f4 Na8 22.b4 Qc7 23.Nxa8 Rfxa8 24.b5 cxb5 25.Bxb5 Ba6 26.c6 Bxb5 27.Qxb5 a4 28.Rc2 Ra5 29.Qb4 R5a6 30.Qd6 Rc8 31.Rfc1 Rb6 32.g4 Qa7 33.f5 exf5 34.gxf5 Qc7 35.Qxd5 Rd8 36.Qe4 Re8 37.e6 fxe6 38.fxg6 h6 39.Qe3 Rb5 40.Qxh6 Re7 41.Rf2 Qd8 42.g7 Rxg7 43.Qxe6+ Kh7 44.c7 [FEN] Rxc7?? 45.Rxc7+ 1-0
The first step is to answer the question "what is the problem?". Qa8 would be mate if it wasn't prevented by some pesky defenders. The next step is "how to get rid of the defenders?". These questions and follow up questions lead to the solution. Both the questions, the scenarios and the patterns should come up naturally.
DeleteThat defines the goal of the training. Only when all these elements are absorbed, you are good to go on.
The first step is seeing the mate in 1 (Qa8+).
DeleteThe second step is investigating the LoAs (lines of attack). Asking the question "How can white frustrate the LoA?" Which points out the defenders which can interfere.
The patterns and the logic should go hand in hand.
In NM Dan Heisman's intriguing book The Improving Chess Thinker, he provides a generic thought process. [I put this in a comment on 24 JAN 206. Obviously, I haven’t burned that process into System 1.]
DeleteThe step-by-step process is:
1. What are ALL the things my opponent's [last] move does?
2. What are ALL the positive things I want to do?
3. What are ALL the candidate moves which might accomplish one or more of those goals?
4. Which of those initial candidates can I reject immediately because they are not safe?
5. Of the final candidates, which one is the best I can find in a reasonable amount of time?
After reading your comments regarding the “first step,” I had an insight (I think) regarding both players’ mindsets, starting with the position following Black’s 43… Kh7. I’m reasonably sure that both players “SAW” the WPc6. It (obviously) blocks any attack from a8 on the White king. I think the problem was that both players had a “retained image” of WPc6 as they contemplated their next few moves.
White felt it was “safe” to advance the pawn, primarily because it was a forcing move, attacking the Black queen. He overlooked that it opened the line of attack for the Black queen on a8.
Black, in his turn, also had the retained image of WPc6, and felt he had to either retreat the queen or just capture the attacking pawn. Since he was behind in material, he may have seen the capture as a way to accomplish two goals: regain a pawn, and getting closer to a drawn queen endgame after an exchange of a pair of rooks.
Both players overlooked one of the crucial things that would have been revealed if they had followed Heisman’s first step. It is necessary (but insufficient) to “SEE” all the POSITIVE things that a move does. We also have to “SEE” all the NEGATIVE things that a move does. Often, it is much easier to identify the “positives” without realizing all the negatives. Apparently, neither player “SAW” the negatives.
This kind of insight is the reason that I try to follow your advice to absorb all the elements—the cues, the patterns, the expectancies and the potential actions. Without your advice (and going back to look at ALL the things that occurred as a result of 47. c7, I would never have had this insight. This is why I much prefer to have the game score, so I can go over the moves leading up to and after a denouement. I like to think about WHY a combination works or not, and try to identify all the surface features and the essence of the position.
Obviously, the comment was dated 2016, not 206. I'm not quite that old.
DeleteGood points. I'm pretty sure you are right.
DeleteSince we are no computers, we cannot complete the tasks Heisman sets out for us within a reasonable time, let alone within the time limits of a game.
Heisman's first point could potentially be used to prune the thought process. That might be worth investigating.
It consists of two points:
1a: what does the piece do on its new square?
1b: what does the piece no longer do by leaving the old square?
How absolute can we take these two points? Might anything emerge in the position that isn't covered by these points?
Repeating that first step:
DeleteWhat are ALL the things my opponent's [last] move does?
Your formulation of two sub-rules makes explicit an assumption about the rule. That is, to look at the moved piece and determine its Function before and after the move. I think the rule covers a broader area than just the moved piece. It also encompasses other pieces (perhaps for both players on both sides of the board) as well. For instance, in the example position from White’s perspective, the advancement of WPc6 increases its proximity to promotion and “attacks” the Black queen. From Black’s perspective, it opens the a8-h1 diagonal as a potential line of attack. I think covers the move according to your sub-rules, restricting them to just the piece moving/moved.
However, after reflection, I think it is broader than that. Let’s start with some surface-level features of the position. The White king is immobilized by the WPh2 and the BRg7. Consequently, g1 and g2 are contested; g2 is B.A.D. but that is unimportant at present. The b1 square is B.A.D.; that also is unimportant PRIOR TO 44. c7; neither player has a way to take advantage of that. The Black king is (relatively) safe at the moment, although potentially subject to attack. On the other hand, the back rank of the White king is “tender” (vulnerable) because of the BRg7 and the immobility of the White king, but Black has no easy way to increase pressure on the b1 PoP. Consequently, the position is in static equilibrium.
The advance of the WPc6 with 44. c7 changes the dynamic balance. The WPc7 is still protected by the WRc1, which has increased its range of action. Unfortunately for White, his pieces do not coordinate/cooperate with each other in mutual defense. The new open LoA appears to be safeguarded by the scattered White pieces: all squares EXCEPT a8 and b7 are guarded by one or more White pieces. What has changed is that Black can now force the WRc1 to be overloaded by having to protect a defender of one of the squares on the long diagonal. After 44… Qa8+, the only possible effective (forced) move (45. Rf3 is useless) to block that LoA is 45. Qc6. This adds a defensive Function that White cannot carry out because of the requirement for WRc1 to protect against the back rank mate. This is the dynamic consequence of advancing the WPc6. After 45. Qc6 Qxc6+ 46. Rxc6 (forced) Rb1+ 47. Rc1 Rxc1#. The advance of the WPc6 dynamically changed the functional capabilities of WRc1. That is not obvious if attention is restricted to the moving/moved piece.
I may have miss-read your intention regarding the restriction of attention solely to the moving/moved piece. If so, my apology.
As for the rest of Heisman’s thinking process, I agree that it is essentially useless for anyone more advanced than a novice. It’s also not particularly useful for a novice either because it is to vague and all-encompassing, but some guidance is necessary when one has little or no practical experience to guide the decision-making process.
Correction: the g2 square is NOT B.A.D.: it is protected by the WKh1 and WRf2. Writing while not looking at the board creates illusions. Sorry about that.
DeleteI look for ways to prune the method of Heisman, in the hope to find a method to prune my own thinking process.
ReplyDeleteWhat does a moving piece possibly do?
It attacks
It comes closer to promotion
It opens LoAs
It closes LoAs
Et cetera
Summarizing:
It opens possibilities
It closes possibilities
All these changes in possibilities are caused by that one move.
If the position is balanced, and one piece moves, the position can change in one of three possible ways:
The position becomes better
The position becomes worse
The position stays the same.
But all changes are related to that one move. Hence, if I take it as the beginning of my considerations, I will ALWAYS find the new possibilities that arise. Pieces that don't move, cannot CHANGE the possibilities.
That gives me a STARTING POINT that is always there. And in an implicit way, it tells me I can prune all other possible starting points without missing something.
The move c7 opens the long diagonal against the white king. If I study that LoA, I will discover all ways that I can use that LoA, and all ways that white can interfere with that LoA. From there, the logic narrative starts to enfold itself.
The title of this post is there for a reason. It hints at the desirability to make it a habit (absorb) to SEE what a move does, (and what it does not anymore). If we can cultivate that habit AND be sure that we don't miss something, we might develop a mighty weapon.
Smirnov reconsiders his positional plans whenever a pawn moves or a piece is traded. Because that are the only long term alterations of a position.
ReplyDeleteSo maybe we can divide the judgment as follows: when a pawn moves, we must judge the change in the LoA landscape, and when a piece moves, we must judge the consequences for the Fun-ctions.
FEN: 8/7p/3p2p1/2pP1kP1/2K4P/pP6/P7/8 w - - 0 54
ReplyDeleteHere is a game from a lichess.org puzzle “#bTatD”. Not best play at several points in this game. For example, after 49. Kxe2 Black had the opportunity to grab the b3 pawn with a stock tactical pattern 49… Bxb3 because if the bishop is captured, Black will promote the a3-pawn. GM Stockfish rates the resulting position with 4.5 for Black.
You can’t execute what you can’t “SEE”!
In the game, White played 54. Kb5 with an evaluation of 0.0 by GM Stockfish. Material is even, both sides have three pawn islands and both sides can defend the central pawns—but only if they “know” a typical scenario/pattern.
The interesting aspect (for me) is what is “hidden” below the surface (or in the not-so-distant future).
Can you “SEE” what Black failed to see when he played 54… Ke5?
If you can, you also “know” how to play out the scenario all the way to checkmate. Not move-by-move, but knowing typical actions to take at each stage regardless how Black wiggles.
[White "Olann"]
[Black "ss4000"]
[WhiteElo "1826"]
[BlackElo "1831"]
[Date "2022.3.??"]
[Result "1-0"]
[Event "5+3 Rated Bitz"]
[Site "lichess.org"]
1. e4 e5 2. Nf3 Nc6 3. d4 exd4 4. Nxd4 Qf6 5. Nxc6 Bc5 6. Qe2 Qxc6 7. Nc3 Nf6 8. Bd2 O-O 9. O-O-O d6 10. f3 a5 11. g4 a4 12. h4 a3 13. b3 Bb4 14. Qe3 Bc5 15. Qd3 Be6 16. Qb5 Nd7 17. Qxc6 bxc6 18. Ne2 Ne5 19. Bg2 Rfb8 20. Bc3 Be3+ 21. Bd2 Bc5 22. Ng3 Rb7 23. Nf5 Kf8 24. Ne3 Rab8 25. g5 Nc4 26. Nxc4 Bxc4 27. Bf1 Be6 28. Bc3 Kg8 29. Bd3 Kf8 30. Kd2 Kg8 31. Ke2 Bd7 32. Rhg1 Be6 33. Rg2 Bh3 34. Rg3 Be6 35. f4 g6 36. f5 Bd7 37. fxg6 fxg6 38. e5 Rf8 39. exd6 cxd6 40. Rf1 Rbb8 41. Rxf8+ Rxf8 42. Bf6 Re8+ 43. Kd2 Be6 44. c3 Bd5 45. Bd4 Bxd4 46. cxd4 Rf8 47. Re3 Rf2+ 48. Re2 Rxe2 49. Kxe2 Kf7 50. Bc4 Ke6 51. Kd3 Bxc4+ 52. Kxc4 Kf5 53. d5 c5 54. Kb5 Ke5 55. Kc6 1-0
A situational assessment should include those aspects of the position that are readily visible, providing “cues” as to what strategy (scenarios) may be applicable.
DeleteIn the given endgame, the broad outlines appear in the three regions of the board: kingside, center and queenside. On the kingside, the pawn structure is statically and dynamically balanced. One (perhaps) crucial aspect is that White’s pawns on the kingside are more advanced. Without trying to figure out the exact number of tempi required to capture the White pawns and then advance a Black kingside pawn to promotion, it should be intuitively obvious that it will require several moves to capture both pawns, and then get a pawn to promotion.
The center pawn situation is more fluid. If Black can eliminate the d5-pawn, then he will have one or two pawns to advance to promotion. The central question will around whether one player or the other can capture or defend that d5-pawn.
The queenside pawn structure is advantageous for White in at least two ways. First, the WPb3-pawn is mobile. It can be used to attack the BPc5, perhaps with the goal of eliminating the two Black central pawns (or at least eliminating the BPc5 through exchange), thus paving the way for the WPd5 to be advanced toward promotion. Second, advancing the b3-pawn eliminates the most significant obstacle preventing the White king from getting the a3-pawn off the board, making the a2-pawn a passer. The question (which can only be answered by calculation—mental simulation with action scripts) is what action(s) can Black undertake while White is eliminating the a3-pawn.
Some of the background scenarios that may prove useful (depending on the actual moves played) would be (generally) Zugzwang of all types (the Trébuchet is almost incidental aspect of the position prior to Black’s blunder and should not have occurred on the board), Silman’s Fox in the Chicken Coop (either White or Black can try to use this idea), the relative value of an outside passed pawn, the square of the pawn, and all sorts of tempo-gaining moves (if/when available). None of these scenarios are specifically in play at this point, but may become significant in the future.
It was only after spending time looking at the position (without trying to force a particular direction to my thoughts) that these general aspects began to appear as cues for scenario patterns, providing potential direction for future play.
This indicates exactly at what point I'm bad at endgames. After 23 year tactics, gambits and attacks, I have no experience in making a plan for the endgame. It is not so much a matter of technique, it is a matter of lacking a plan. Or at least having some overview on what is going on in the position.
DeleteIt should be not too hard to fix, I guess. Hope.
Two things have helped me: (1) Taking GM Aagaard's advice to "SLOW DOWN! and (2) allowing time to let the cues pointing toward patterns "speak" to me (through System 1) from the viewpoint of the specific position, unforced by my preconceived notions. After the cues become more visible, then the corresponding patterns begin to come into consciousness, giving guidance toward a plan. A plan is NOT for the rest of the game (far too many unknowns to do that) but merely to reach a point where the next scenario can be seen. After implementation of that scenario is completed, lather, rinse, repeat.
DeleteEasy to say, but not easy to do. Sigh.
The advice to “slow down” is very important, especially during training. Too often, I have found myself “solving” a problem and then moving on, instead of taking the time to determine the before-and-after context for what happened, trying my best to uncover as many “secrets” buried (perhaps slightly) below the surface. When I initially “SAW” the Trébuchet, I was only surprised that the Black player did not see it; I didn’t consider it to be the leitmotif of the position, or really important because it should have been relegated to a side note at most. However, it was not until I contemplated what was going on over the entire board that I gained some insight into the surface cues pointing to possible plans.
Let me guess. A trebuchet?
ReplyDeleteOf course. It's also a reasonably simple example of how experience/skill can guide the realization of scenarios through mental simulation and action scripts without having to know every individual move in the ensuing process. The process moves from one scenario to the next IFF we have all of the required scenarios in long-term memory. Absent that basis in skill, we are reduced to being novices and must revert to a step-by-step logical process in order to calculate our way forward to an intermediate result.
DeleteThe lichess problem occurred after 54. Kb5. White set a trap for Black. Black (obviously) did NOT know the trebuchet pattern or even generally how to avoid specific mined squares in any “simple” endgame. Given that this was a blitz game, I hazard a guess that Black did not learn anything from this experience that could be used in the future.
Nothing to “SEE” here, it’s “just” a trebuchet; on to the next game!
Logically, the game should have ended in a draw from that problem point. After I found the game score, I went through it from move one and was surprised to find so many errors by both players. I had assumed (wrongly) that given the players' ratings, there would not be so many errors. Maybe an error or two at the end, but not through the late opening and entire middle game. My goal in working through the game score is to identify the point at which a specific pattern can first be identified (even if obscured), and then to identify as many cues as possible at that point which hint at that elusive future pattern.
Skill is acquired through experiences that are analyzed in depth for cues pointing to patterns with associated actions, i.e., scenarios. Without the analysis with focused attention, mere experience is simply wasted.
I recall something GM Nigel Davies said:
It really doesn't matter what you study, THE IMPORTANT THING IS TO USE THIS AS A TRAINING GROUND FOR THINKING [N-O-T FOLLOWING A STEP-BY-STEP THINKING PROCESS BUT DOING THINKING—ABOUT WHAT IS RIGHT IN FRONT OF YOUR EYES!] RATHER THAN TRYING TO ASSIMILATE A MIND-NUMBING AMOUNT OF INFORMATION.
PART I:
ReplyDeleteNOT a Trébuchet!
FEN: 2n1k1r1/p7/3B1Rp1/2P2pKp/8/4P1P1/5P1P/8 w - - 17 45
The given position occurred in a computer match back in the Dark Ages before the rise of the Chess Terminators. It’s a position that can be evaluated by intuitive valuations of the various salient aspects. What are the salient factors to consider in this position?
Should White play 45. Rxg6 or not?
First scenario: capture the g6-pawn and exchange Rooks. Evaluation: two pawns up in material with the possibility of capturing at least one or more kingside pawns with the greatly enhanced possibility of promoting a White pawn.
Second scenario: Black gets a turn. Capturing the WBd6 eliminates any possibility of the long-range Bishop dominating the Black Knight and perhaps being able to prevent Black’s a7-pawn from promoting. As soon as that Bishop is gone, Black can advance his a7-pawn two squares (after 45… Rxg6+ 46. Kxg6 Nxd6 47 cxd6). Wrinkle #1: Black does NOT have to recapture the new WPd6 immediately after 46...Nxd6! The Black King is in the square of that pawn.
Third scenario: can the White King rush over to stop the Black a-pawn after 47… a5? No, the White King is outside the “square of the pawn” (mental model/”rule”). White also cannot eliminate one of the remaining Black kingside pawns, and get out of his own way to get his own passer to promotion before or simultaneously with Black’s promotion.
Conclusive evaluation: capturing with 45. Rxg6 is NOT a good idea!
At this point, option 1 can be discarded; there is no need for a comparison to any other options—this particular option is toxic.
The amusing aspect of this story is why Nuchess did not “SEE” the refutation of this option.
William Blanchard of the Nuchess team dejectedly confirmed that this program failed to include the principle of the King having to be inside the square of the pawn to catch it. Of course, Nuchess’s search did not extend so deeply that it could see the a-pawn queening. Hyatt said that Cray knew about the square of the pawn rule, but that this was the first time he was aware the principle was relevant to the computer’s move choice…
If the patterns (principles/”rules”/whatever) are embedded in LTM, then it is highly likely that we will NOT “SEE” the available options.
There are several other good options that GM Stockfish identifies. Each of the alternative options should be assessed using the same process: treat each alternative as an option and work out the scenarios associated with it. Recalling GM Davies training advice:
The reality is that YOU'VE GOT TO MOVE THE PIECES AROUND THE BOARD AND PLAY WITH THE POSITION. Who does that? Amateurs don't, GMs do…
The following concrete lines are given by GM Stockfish after about 5 minutes of analysis.
D27 +5.28 45.c6 Nxd6 46.Rxd6 Rg7 47.Rd5 h4 48.gxh4 Rc7 49.Rc5 a5 50.Rxa5 Rxc6 51.h5 gxh5 52.Rxf5 h4 53.Kxh4 Rh6+ 54.Rh5 Re6 55.Rg5 Kf7 56.Kg3 Kf6 57.h4 Re4 58.Rg4 Re8 59.Kf4 Re5
D27 +5.27 45.h4 a5 46.Be5 a4 47.Ra6 Kf7 48.Rxa4 Ke6 49.Bd4 Ne7 50.Ra6+ Kd5 51.Rd6+ Kc4 52.Bf6 Nd5 53.c6 Kc5 54.Rd8 Rxd8 55.Bxd8 Kxc6 56.Kxg6 Kd7 57.Bf6 f4
D27 +5.23 45.f4 a5 46.Be5 Ke7 47.Ra6 Re8 48.Rxa5 Ke6 49.c6 Nd6 50.Bxd6 Kxd6 51.Kxg6 Kxc6 52.Re5 Ra8 53.Kxh5 Rh8+ 54.Kg6 Rh3 55.Rxf5 Rh8 56.e4 Re8 57.e5 Rg8+ 58.Kh5 Rh8+ 59.Kg4
D27 +5.19 45.Be5 Ne7 46.c6 Nc8 47.c7 Kd7 48.Ra6 Re8 49.f4 Ke7 50.Kxg6 Rf8 51.Kh6 Re8 52.h4 Kd7 53.Kxh5
D27 +5.18 45.Bf4 a5 46.Ra6 h4 47.gxh4 Kf7 48.Be5 a4 49.Rxa4 Re8 50.Ra8 Rxe5 51.Rxc8 Re4 52.Rc7+ Ke6 53.Rc6+ Kd5 54.Rxg6 Rg4+ 55.Kxf5 Rxh4 56.Rg5 Kxc5 57.f4 Rxh2 58.e4 Kd4 59.e5 Kd5 60.Kf6 Rh6+ 61.Rg6
ERRATA:
DeleteIf the patterns (principles/”rules”/whatever) are NOT embedded in LTM, then it is highly likely that we will NOT “SEE” the available options.
PART II:
ReplyDeleteD27 +5.15 45.f3 Nxd6 46.Rxd6 Rg7 47.f4 Re7 48.Kxg6 Rxe3 49.Kxf5 Rc3 50.Ra6 Rxc5+ 51.Kg6 Rc2 52.h3 Rg2 53.Ra3 Rh2 54.h4 Rc2 55.Kxh5 Rc7 56.Kg4 Rb7 57.h5 Kf7 58.Ra6 Kg8 59.Kh3 Kh7 60.g4
D27 +5.12 45.Bc7 Rf8 46.Ra6 f4 47.Bxf4 Rf5+ 48.Kxg6 Rxc5 49.Bg5 Rc2 50.Kxh5 Rxf2 51.h4 Rg2 52.g4 Rb2 53.Kg6 Rg2 54.h5 Rxg4 55.h6 Rg1 56.h7 Rh1 57.Kg7 Rh3 58.h8=Q+ Rxh8 59.Kxh8
D27 +5.11 45.h3 Kd8 46.Be5 Re8 47.f4 a5 48.Ra6 Ke7 49.Rxa5 Ke6 50.c6 Nd6 51.Bxd6 Kxd6 52.Re5 Rg8 53.c7 Kxc7 54.h4 Kb6 55.Kf6 Rd8 56.Kxg6
D27 +5.07 45.Bb8 Rf8 46.Ra6 f4 47.Bxf4 Rf5+ 48.Kxg6 Rxc5 49.Bg5 Rc2 50.Kxh5 Rxf2 51.h4 Rg2 52.g4 Rb2 53.Bf6 Rb6 54.Rxb6 Nxb6 55.Kg6 Nd7 56.Bb2 Nf8+ 57.Kg7 Ne6+ 58.Kg8 Kd7
D27 +4.91 45.Kh6 Kd7 46.Bf4 a5 47.h4 Re8 48.Ra6 Ne7 49.Rxa5 Kc6 50.Bd6 Nd5 51.Kxg6 Nc3 52.Kf7 Rd8 53.f3 Nb5 54.Ra6+ Kd5 55.Bf8 Rd7+ 56.Kg6 Rd8 57.Rf6 Kc4 58.Rxf5 Kd3
D27 +4.59 45.Kh4 g5+ 46.Kxh5 a5 47.Be5 Rh8+ 48.Kxg5 Rxh2 49.Rxf5 a4 50.Rf6 a3 51.Ra6 Rxf2 52.Rxa3 Kd7 53.Ra6 Ne7 54.Rd6+ Kc8 55.g4 Rf1 56.e4 Rc1 57.Kf6 Ng8+ 58.Kf7
D27 +4.53 45.Re6+ Kd7 46.Rf6 Kc6 47.Bf8+ Kd5 48.c6 h4 49.gxh4 a5 50.c7 f4 51.exf4 Kc4 52.Ba3 Rg7 53.Rc6+ Kb5 54.Kh6 Rd7 55.Rc2 Rd5 56.Kxg6 Rd3 57.Bf8 Rf3 58.h5 Rxf4
D27 +4.52 45.Kf4 g5+ 46.Kxf5 Kd7 47.Be5 Ne7+ 48.Ke4 Rc8 49.Bd6 Nc6 50.Kd5 Nb4+ 51.Kc4 a5 52.Rh6 h4 53.Kb5 Nd5 54.Be5 Ne7 55.gxh4 gxh4 56.f4 a4 57.Rd6+ Ke8 58.h3
D27 +4.34 45.e4 fxe4 46.c6 Kd8 47.Bf4 Ke8 48.c7 Kd7 49.Ra6 Rf8 50.Rxg6 Ne7 51.Rd6+ Kc8 52.Kxh5 Re8 53.h4 Kxc7 54.Kg4 Kb7 55.Re6 a5 56.Rxe4 Rg8+ 57.Kh3
D27 +3.83 45.g4 hxg4 46.c6 Kd8 47.Bf4 Ke7 48.c7 Nb6 49.Rd6 Nc4 50.Rd5 Nb6 51.Ra5 Nc8 52.Ra6 Kd7 53.Bg3 Re8 54.Rxg6 Ne7 55.Ra6 Rg8+ 56.Kh6 Re8 57.Kg7
D27 +0.23 45.Bf8 Rxf8 46.Rxg6 Ne7 47.Ra6 h4 48.Kxh4 Rh8+ 49.Kg5 Rxh2 50.Rxa7 Kf7 51.f3 Rg2 52.g4 fxg4 53.fxg4 Rc2 54.e4 Rxc5+ 55.Kh4 Rc4 56.e5 Rc8 57.g5 Rd8 58.Rc7
D27 0.00 45.Be7 Kxe7 46.h4 Ke8 47.Ra6 Rg7 48.Rc6 Ne7 49.Rc7 Kf8 50.Rxa7 Rf7 51.Ra4 Kg7 52.Ra7 Kf8
D26 -5.10 45.Rxg6 Rxg6+ 46.Kxg6 Nxd6 47.e4 a5 48.exf5 Nxf5 49.Kxf5 a4 50.Kg6 a3 51.Kxh5 a2 52.Kg4 a1=Q 53.h4 Qd1+ 54.Kf4 Qe2 55.c6 Qc4+ 56.Ke3 Qc5+ 57.Ke2 Qxc6 58.h5 Qe4+ 59.Kf1 Kd7 60.Kg1 Kd6 61.h6 Ke5 62.f4+ Kd4 63.h7 Qxh7
D26 -5.55 45.Rf8+ Rxf8 46.Bc7 Kf7 47.h3 Ne7 48.g4 Rc8 49.Bd6 Nc6 50.f4 hxg4 51.hxg4 a5 52.gxf5 gxf5 53.Kxf5 a4 54.Ke4 a3 55.Kd3 a2 56.Kc2 a1=R 57.Kb2 Rf1 58.f5 Rxf5
D26 -5.91 45.Rf7 Kxf7 46.Be5 Ne7 47.f3 Nc6 48.Bc3 Rd8 49.e4 fxe4 50.fxe4 Rd3 51.Bb2 Re3 52.Bf6 Rf3 53.Ba1 a5 54.Kh6 a4 55.Kh7
Game score:
[White "Nuchess"]
[Black "Cray Blitz"]
[WhiteElo "2100"]
[BlackElo "2300"]
[Date "1984.10.09"]
[Result "0-1"]
[Event "ACM Computer Championship"]
[Site “San Francisco, California”]
1.c4 e5 2.Nc3 Bb4 3.a3 Bxc3 4.dxc3 Ne7 5.g3 d5 6.cxd5 Qxd5 7.Qxd5 Nxd5 8.Bg2 Nb6 9.a4 O-O 10.a5 Nc4 11.Ra4 Nd6 12.a6 Nd7 13.Be3
Nb6 14.Rh4 Rd8 15.axb7 Bxb7 16.Bxb7 Nxb7 17.Nf3 Rd5 18.c4 Ra5 19.O-O Ra2 20.Rd1 Rxb2 21.c5 Nc8 22.Rd7 f6 23.Rg4 g6 24.Rh4 h5
25.Rxc7 Nd8 26.Ra4 Rb7 27.Rxb7 Nxb7 28.Ra6 Kf7 29.Nd2 Nd8 30.Ne4 f5 31.Ng5+ Kg7 32.Nf3 Nf7 33.Nxe5 Nxe5 34.Bd4 Kg8 35.Bxe5 Ne7
36.e3 Kf7 37.Rf6+ Kg8 38.Kg2 Rc8 39.Kf3 Re8 40.Ra6 Ra8 41.Kf4 Kf7 42.Kg5 Rg8 43.Rf6+ Ke8 44.Bd6 Nc8 45.Rxg6 Rxg6+ 46.Kxg6 Nxd6
47.cxd6 a5 48.g4 hxg4 49.Kxf5 a4 50.e4 a3 1.Kxg4 a2 52.e5 a1=Q 53.f4 Qg1+ 54.Kf5 Qxh2 55.e6 Qc2+ 0-1
PART III:
ReplyDeleteAny given position may be a complex, ambiguous set of tangled cues. When a task is ill-structured, there are typically several equally good ways of solving the same problem. There is no one accepted procedure to use, and it is necessary to select or invent a way to proceed. Moreover, there is no single correct or best answer. Ill-structured problems frequently are made more ambiguous by uncertain dynamic information and by multiple interacting goals.
In light of the inadequacy of the first option (chasing immediate material gain, only to discover that the opponent gets the last laugh by promoting his outside passed pawn), what other plan(s) could White investigate?
I included 20 variations for a specific reason: I wanted to capture GM Stockfish’s evaluation of option 1 (45. Rxg6). As can be seen, Black could still mess it up by, for instance, playing this line: 45.Rxg6 Rxg6+ 46.Kxg6 Nxd6 47. cxd6 Kd7 [going after the d-pawn as material gain, which is unimportant compared to getting the a-pawn moving as quickly as possible] 48. Kxf5 and White will be within the square of the a-pawn; Black will lose.
I recall GM Aagaard’s advice to SLOW DOWN! There are multiple connotations to that advice: taking more time on the clock to carefully evaluate other options, avoid rapid-fire assessments and dismissal of other options, or (perhaps) go back to examining the overall position, incorporating the information discovered by evaluating the scenarios in option 1 as the first variation processed.
Some of what has been uncovered is applicable to other options. White cannot focus on gaining material on the kingside while ignoring that crucial outside passed a-pawn. Some alternatives are: advancing the White pawns so as to create a passed pawn prior to chopping off the g6-pawn, so that White wins the pawn race; retaining the White Bishop so that there is a piece that can prevent the a-pawn from promoting; refocusing on the a-pawn and attempt to exchange it off for the c-pawn, and so forth.
It does NOT matter which option is selected as option 2 to consider! The important thing is to find the first option that retains White’s advantage. As can be seen from GM Stockfish’s analysis, there are 15 options that should be “good enough for government work.” Because those options are “winning,” they constitute a “zone of indifference”; it really doesn’t matter which one is chosen because each one gives White winning chances.