Think like a grandmaster
Back in 2000 I realized that when a 14 year old kid starts to beat up chess players who have 50 year or more experience under their belt, there must be some sort of trick that they apply.
When Susan Polgar set her simul record in 2005, I did the math and came to the conclusion that whatever the trick was, it had to be subconscious. She used 2.6 second per move at average, including walking from one table to another, and there is no way that someone can think in such short amount of time.
Hence I started my blog in 2005. Many red herrings were thrown at me by all kinds of chess authors, and I catched them all.
In 2023 I ran out of red herrings, and I concocted some sort of method to educate my subconscious. It kind of worked, although it was obvious that it was far from efficient.
At May 5th, 2026, finally all pieces of the puzzle fell into their place. In order to obtain the trick, you must absorb the logical patterns of a position, not so much the geometrical patterns. Since logical patterns are transferable from one position to another, while geometrical patterns are not.
So the Trick is, that a grandmaster prunes the tree of analysis rigorously before he even starts to think. Since this happens subconsciously and lightning fast, they are not aware of themselves doing so.
In order to be of service to their students, they invented all kinds of confabulations, and throw them out as red herrings over the chess community.
In order to find out the difference between geometrical patterns and logical patterns, compare the two following diagrams:
 |
| Diagram 1. Black to move |
r3kb1r/pp3ppp/2N1pn2/q7/3P4/2N5/PP2bPPP/R1BQ1RK1 b kq - 0 11
Siebrecht, S. vs. Odendahl, R., Wuppertal 1994
 |
| Diagram 2. Black to move |
r2B2k1/1pqb1pp1/p6p/8/PbB5/1P1p1N1P/2P1QPP1/R5K1 b - - 0 22
Kamsky, G. vs. Akobian, V., St. Louis 2017
Can you spot the logical pattern that these two positions have in common?
Geometrical patterns are the easiest to recognize. Hence everybody does that. It is hard to make a difference there.
ReplyDeleteLogical patterns have a degree of abstraction in them. This makes them much more common. But they are harder to learn. You must isolate them from a puzzle, conceptualize them so that they become position independent. Then you must absorb them. That is much harder to do. But that is how you learn the Trick.
Tempo observed:
DeleteLogical patterns have a degree of abstraction in them.
I presume there is a higher level of abstraction where we enter the realm of heuristics (general principles, "rules", whatever) - and the current [Watson] consensus is that "rules" are largely useless in specific positions because there are too many exceptions to them for them to be useful.
Obviously if we "isolate them [the logical patterns] from a puzzle [or multiple puzzles], conceptualize them so that they become position independent," then we will have reached a fairly high level of abstraction above the geometrical pattern level.
Are there any "signs" which warn us that we are going too far with the abstraction process?
You describe the combination in logical sentences.
DeleteA is defending B. How can I annihilate A? I could do so if it wasn't for C. What can I do about C? Et cetera.
System 2 translates the combination into sentences. System 1 works its miracles on the sentences and make subconscious patterns out of it. That are the logical patterns we are hunting for.
Just like System 1 glues the image of a word to its meaning.
This is the first level of abstraction.
When you do a lot of tit for tat puzzles, you might well notice that in certain positions a queen becomes a desperado and in other positions it does not. Here you might find another level of abstraction (level 2). And System 1 will works its miracles and will help you to recognize a desperado.
But every level of abstraction leaves a lot of details in the dust. I guess you will notice when you go too far.
“In desperate position, you must fight.” – Sun Tzu
ReplyDeleteDr. Lasker's motifs come in handy for encapsulating common and important logical patterns in a single word.
"When you absolutely don't know what to do anymore, it is time to panic." GM John van der Wiel
ReplyDeleteOut of curiosity, I switched the side to move in both positions. (I have a perverse mind.) In both cases, White was winning with extra material. Is "first player to move has a winning sequence" important? YES, but in both these cases, it is Black to move, so White doesn't get a chance.
ReplyDeleteIt are both tit for tat combinations where all of a sudden the queen goes haywire
DeleteI "SAW" the queen go "desperado" before I realized it was a "tit for tat" combination. I'm curious: did you recognize the position as tit for tat before you "SAW" "all of a sudden the queen goes haywire"? Or did the recognition of both patterns occur simultaneously?
DeleteThe puzzle position starts a few ply earlier. And that looked as a tit for tat position. I'm not a great desperado recognizer. But that is not what inspired this post.
DeleteI'm making progress. But it is too early to draw conclusions.
ReplyDeleteI'm pretty euphoric that I finally begin to make use of my usual logic. Unbelievable that it took me so long to dispel my mental confusion.
Tempo observed:
ReplyDeleteIn order to obtain the trick, you must absorb the logical patterns of a position, not so much the geometrical patterns. Since logical patterns are transferable from one position to another, while geometrical patterns are not.
I presume (please correct me if I am wrong as usual) that there is some way to distinguish between geometrical patterns and logical patterns.
Are the typical named tactical devices/themes/motifs the same as the components of the geometrical patterns? If so, then what are the components of the logical patterns category? Are there any training materials (guidance) for learning the components of this more abstract category?
A geometrical pattern connects the piece position to an inner structure.
ReplyDeleteA logical pattern connects the logic to an inner structure.
Both is done by System 1