How good is a grandmaster at tactics?
I concocted my new method for transfer of knowledge from system 2 to the patterns of skill in system 1 in March 2023. So far I have absorbed 520 mates and 100 tactics (non mates) in about 13 months.
GM John Nunn played a series of 100 blitz games against a 2300 rated master. The resulting score was 88-12, mainly based on LPDO. Nunn was rated about 2600 back then.
Which leads to the conclusion, that even between high rated players, the mastering of simple tactics still show a huge difference. So what do these high rated players see and not see?
I bought the book and video of Chess for Zebras (hat tip to Robert). In the video GM Jonathan Rowson describes a game of him against GM Viktor Korchnoi. He was rated 2540 back then, while Korchnoi was about more than 100 points higher rated.
Rowson's last move was 11. ... Qf6?
 |
| White (Korchnoi) to move |
r3k2r/pppn1pbp/1nb1pqp1/8/3P4/1QN1BN2/PP2PPPP/3RKB1R w Kkq - 1 12
Korchnoi jounced up with a "this can't be good" frown on his face. And played, after just a minute checking all possibilities, with confidence a move that Rowson totally missed. As if Korchnoi was playing a simul. What did Korchnoi SEE?
Description of Rowson:
"This move is not particularly deep or difficult, yet I was taken aback by the fact that Korchnoi played it within a minute. When I completed the move 11... Qf6, he didn’t laugh or scowl (as he has been known to), but he did furrow his brow as if to say “that can’t be right” and then he looked again for just a few seconds, during which time I realized what was about to hit me. Then it happened. As soon as he saw the move there was not a trace of hesitation that it was the right one. This sense of conviction made a lasting impression and I believe it helped me to win a game later in the same tournament (Rowson-Sarakauskas in Chapter 7) ."
Bd4 wins the Bg7.. ifffff
ReplyDeleteAt RC. So d5 square clearance followed by Bd4 threatens the Queen and wins the bishop.
DeleteAnonymous is not RC this time, Jim.
DeleteThere's a contextual cue along the diagonal: the Black Queen "protects" the Black Bishop and cannot move along a file or rank to continue that protection (because of the two pawns) IFF the Black Queen is attacked along the diagonal. That conditional IFF should trigger a search for a way to create that attack. For instance, when you mentally classify a piece as a LPDO or “defender,” that should immediately trigger a search for potential attackers.
This contextual cue applies in many similar instances involving other pieces, including the King. Extracting and abstracting (ie, generalization of) the essence is a very important part of training so that we can SEE that salient cue whenever it occurs, regardless of the specifics of the concrete position.
If we use "pieces-on-squares" for our cues, then we will most likely "SEE" only the surface-level salient cues and be blind to all hidden salient cues. It is also be very likely that we will NOT find those hidden salient cues through "calculation."
I think it was very “telling” that GM Rowson admitted to SEEing the consequence of GM Korchnoi’s 12. d5! — immediately AFTER it appeared on the board. No cigars for “Woulda, coulda, shoulda”!
Bg5 looks good . . But the queen can move to f5. Can she be trapped ?
ReplyDeleteThanks Rc/ts . An instructive position. I need to create a database of these.
ReplyDeleteAn interesting aside: I looked up the game Korchnoi-Rowson in the Chess Tempo database. There are NO problems associated with this game.
DeleteI was hoping to find at least one problem, so I could see what tags had been applied to it/them. Given there are no problems, it's a crap shoot as to whether similar problems can be found using the tag system.
Does anyone with a Chess Tempo account have any idea what tag(s) might be applicable to this concept? I tried sampling the Skewer, X-Ray, Diagonal Battery, Diagonal Clearance and Diagonal Control tagged problem sets but found no problems that were conceptually similar to the game position after an hour or so of searching.
The material is equal with two bishops for a rook and pawn. So I understand why you will not find it. But positionally white is much better. Apart from the psychological blow.
DeleteI assume the black Q-B position is the first cue to see as being suspect. Then you need to see the LoA B-d4-Q-B skewer. The only defender that can interfere with the LoA is the e-pawn. d5 is a multi purpose move. It both clears the LoA AND lures the e-pawn away from the LoA
ReplyDeleteIt is not rocket science, but you must learn to see it.
I wanted to give an impression. How good is a grandmaster tactically? Good, but not perfect, and not totally out of reach. In the sense that it is not incomprehensible.
ReplyDeleteI think there is a further line. Grandmaster vs Super Grandmaster.
ReplyDeleteOne of the ‘interesting’ aspects of the Korchnoi-Rowson problem is the surface-level salient cue consisting of the static formation BQf6-BBg7-f7-g6, creating a situation which severely limits the movement of the Black Queen to escape an attack while continuing defense of the Black Bishop. Labels are unimportant, but I envision this formation as a “box canyon.” Those familiar with military tactics will recognize the danger posed by putting troops into a narrow ravine that severely limits their mobility in the event of surprise attack (“like shooting fish in a barrel”).
ReplyDeleteOut of curiosity, I got out Maxim Blokh’s Combinative Motifs and did a quick search for problems with similar “box canyon” formations in order to SEE if that might be a usable salient cue in other situations.
The following three problems illustrate the concept. Please keep in mind that I am NOT claiming any special “discovery”; I’m just pointing out the similarity of one of the salient cues even though the squares and the pieces involved are entirely different from Korchnoi-Rowson.
Problem 17:
Difficulty: 1
FEN: 1Q6/2r3pk/6bp/3pq3/p4N2/2P3P1/PP6/K1R5 w - - 0 1
Obvious tactical heuristics spring to mind:
“If your Knight is on the same color square as your opponent’s King and/or Queen and is within one move of forking either of those pieces, then look for a sacrifice to bring the other piece into position for a royal fork.”
“The g6-square is B.A.D. Add an attacker or remove a defender in order to gain material.”
Problem 142:
Difficulty: 6
FEN: r3r1k1/p2q1ppp/2p5/2bp4/5Pn1/1PQ5/PBP3PP/RN3R1K b - - 0 1
“Always check; it might be mate!”
There is a potential back rank issue, but White appears to have everything covered [2:1]. But look at that “box canyon” with the additional punch that the WRa1 is also on that same LoA and is immobilized by the WNb1. Hmmm . . .
Problem 340:
Difficulty: 2
FEN: 8/3b1pk1/p1q2rpp/4Q3/4P2P/BPPR2R1/P5K1/5r2 w - - 0 1
“When an opponent’s piece is pinned, add attackers or remove defenders.”
“When a defender has no freedom of movement, an attack on the defender may divert it from its defensive function.”
Not exactly what we train to SEE as tactical cues, is it?
Working on tactics mainly through Ct-art and the 5x5 mainly I have developed some labels for kinds of problems. It is fun and helpful to quickly categorize them and recognize. I'd call this a dovetail formation but I see the power in yours.Some I think as the power of two knights, some are piling on pins,reloaders etc. The heuristic that plays itself out a lot is if a bishop is able to support a Rook on the eighth rank despite multiple pieces in the way move heaven and earth to make it happen. It is often mate.will check out the Blokh problems. Thanks
ReplyDeleteChess Tempo defines two variations for the Dovetail Mate and the Dovetail Mate with Bishop. The distinction between them is simple: instead of two obstructions by the opponent’s side of the two squares that are a Knight’s distance from the focal point, a Bishop controls those two squares along a diagonal. In both cases, the checking piece (Queen or Bishop) must be protected if it is get “up close and personal” to the King and the checkmate occurs with the Queen or Bishop checking the King along the diagonal.
ReplyDeleteExample 1 [with Queen as the mating piece]:
[FEN "2kN4/4bp2/p5pp/1p6/4q1P1/2KR4/P1P2P1P/1R6 w - - 0 1"]
1.Rbd1 Qb4# 0-1
If you “imagine” a WBb2 and WPb3, then either the Queen or the Bishop can mate on b4.
Example 2 [with Bishop controlling the two squares]:
[FEN "rR6/4bk2/2p3q1/4QpN1/2PB1Pb1/4P3/r5R1/6K1 b - - 0 1"]
1...Bxg5 2.Qe8# 1-0
The Epaulette Mate, on the other hand, occurs along vertical and horizontal LoA. If the opponent’s King is at the edge of the board (or hemmed in by its own pieces), a Queen or Rook can deliver the mate without being supported by another same-side piece/pawn.
Example 1:
[FEN "6k1/pp6/8/3p1pp1/r2qp3/KP6/P3QP2/7R w - - 0 1"]
1.bxa4 Qc3# 0-1
Variation with Rooks:
[FEN "6k1/ppr5/1r6/3p1pp1/n3p3/KP6/P3QP2/7R w - - 0 1"]
1.bxa4 Rc3# 0-1
I consider these two Mates to be variations of a theme, not totally separate and distinct patterns.
IMHO, the important thing to do to broaden our capacity to SEE various permutations during actual play is to “fiddle around” with the pieces (or “shuffle the deck chairs on the Titanic”) in order to expose System 1 to as many variations as possible and thereby to burn the possible variations into long-term memory. Very similar “pictures” are linked closely together in long-term memory. Most of us do NOT “fiddle around” with puzzles; we “solve” the puzzle and move right along to the next one, thinking that “quantity trumps quality” (the impetus underlying the Woodpecker approach to the gates of the Seven Circles of Hell). My opinion is that there is much more value in studying one problem position, fiddling around with different configurations until we have unearthed all potentialities. “Name it and claim it as your own.”
The labels we place on any given position are relatively unimportant. If some “word salad” is memorable to YOU, then use it as a trigger. It is the encapsulation of the concepts in a memorable way (perhaps unique to each one of us) that allows System 1 to retrieve the corresponding salient cues whenever we encounter similar situations (or local parts of similarity) in the future.
This was a really valuable game to look at regarding the original point made, i.e. what distinguishes higher levels of mastery from lower levels, so thanks for posting the reference example. After the forced tactical sequence, to the non-master eye Black may look fine on move 18, and the old "points" scale of material balance tells us that as well. Korchnoi however apparently immediately understood that the matchup of the minor pieces vs. rooks was an inherent winning advantage, and modern engines agree. Some other fascinating tactics occur later in the game however, without which White could not win, so this was hardly a one-move blunder of a game.
ReplyDelete