SF burns analysis of GM Kotov on his book How To Think Like A Grandmaster

[MikeGL]

This position was shown in page 20 of Think Like a Grandmaster by GM Kotov.
Don't get me wrong, I am a big fan of this book and of Kotov too. In fact I even read/ analysed this book twice during my teenage days.
The communication during those years when Kotov wrote the book was by snail mail (in postal service) and the variations was
then printed in chess journals or newspapers chess section.

Flohr-Fine 1935/6
[d]r1r3k1/p4ppp/bpq1Nn2/4R3/1PBR4/2Q1PP2/6PP/6K1 w - - 0 1
Here, Flohr made a losing move 24.Nd8? which made R. Fine win this game.
Quote from the book:
"Annotators the whole world over analysed this position. A win for White found in one country was quickly refuted in articles published in another. A practically invisible finesse spotted by one analyst was soon shown to be an error on further examination. Finally the English master Winter found the one and only way to win."

...one and only way to win... (24.b5!)

The funny thing is that SF+NNUE said 24.Nxg7! is even stronger than 24.b5! which points out that Kotov is wrong when he said "one and only way to win with 24.b5".
Ok, to be fair with GM Kotov 24.b5 is 2nd best line for SF-NNUE on my very old machine.

Another wrong claim on this page, where SF found a refutation,
"24.b5 Bxb5 25.Nxg7 Bxc4 26.Nf5! Qa4, and after Black has run out of checks he has no defence against Rg8+ (+ -)"

Again, this line is drawish according to SF+NNUE and not an easy win for white.
[pgn]
[Event "Computer chess game"]
[Site "Talkchess"]
[Date "2020.09.15"]
[Round "?"]
[White "S. Flohr"]
[Black "R. Fine"]
[Result "*"]
[BlackElo "2400"]
[SetUp "1"]
[FEN "r1r3k1/p4ppp/bpq1Nn2/4R3/1PBR4/2Q1PP2/6PP/6K1 w - - 0 1"]
[Termination "unterminated"]
[PlyCount "24"]
[WhiteType "human"]
[BlackType "human"]

1. b5 Bxb5 2. Nxg7 Bxc4 3. Nf5 Qa4 4. Re8+ Rxe8 5. Rg4+ Kf8 6. Qxf6 Qd1+ 7.
Kf2 Qc2+ 8. Kg3 Qxf5 9. Qxf5 Be6 10. Qxh7 Bxg4 11. fxg4 Rxe3+ 12. Kf2 Rae8
[/pgn]

[AdminX]

Hi Mike,

I tested this on my far weaker tablet (i5 Google Pixel Slate) with SF12 (Android Emulation) and it also spots 24. Nxg7 in no time.

[Jouni]

All engines I tried play Nxg7 instantly. Even some 1990s analysis are totally busted with fast computers and/or tablebases. My feeling is that chess is practically solved now .

[Guenther]

Hi Mike,

I tested this on my far weaker tablet (i5 Google Pixel Slate) with SF12 (Android Emulation) and it also spots 24. Nxg7 in no time.

Something is wrong here.
Your second line actually would simply transpose to the first line with 3. Rxb5 instead of 3. Rg5+

[AdminX]

Hi Mike,

I tested this on my far weaker tablet (i5 Google Pixel Slate) with SF12 (Android Emulation) and it also spots 24. Nxg7 in no time.

Something is wrong here.
Your second line actually would simply transpose to the first line with 3. Rxb5 instead of 3. Rg5+

I don't think anything is wrong, this was just a quick test running in a much slower environment. I believe if I let it run longer the quality of the output would be much better.

[Milton]

Here is a fairly deep analysis of the position in which Stockfish never deviates from Nxg7.

r1r3k1/p4ppp/bpq1Nn2/4R3/1PBR4/2Q1PP2/6PP/6K1 w - - 0 1

Analysis by Stockfish 080920:

1.Nxg7 Bxc4 2.Nf5 Kf8 3.Rd6 Qxd6 4.Nxd6 Be6 5.Qa1 Ke7 6.Nf5+ Ke8 7.h3 h6 8.Qd4 Rd8 9.Qh4 Ng8 10.Ng7+ Kd7 11.Nxe6 fxe6 12.Qf4 Re8 13.Re4 Ne7 14.Qf7 e5 15.Rxe5 Rac8 16.Kh2 Kd8 17.Re4 b5 18.Qe6 Kc7 19.Qe5+ Kb7 20.Rd4 Nc6 21.Qxb5+ Kc7 22.Rc4 Re6 23.Qf5 Rd6 24.Qe5 a6 25.b5 axb5 26.Qxb5 Kd8 27.Rxc6 Rcxc6 28.Qa5+ Kd7 29.Qf5+ Re6 30.e4 Ke8 31.Qh5+ Kf8 32.Qb5 Ke7 33.Qb7+ Kf6 34.f4 Rxe4 35.Qxc6+ Kf7 36.Qxe4
White is clearly winning: +- (8.17) Depth: 57/99 13:47:21 1949579MN, tb=2147483647
(, 15.09.2020)

[mwyoung]

All engines I tried play Nxg7 instantly. Even some 1990s analysis are totally busted with fast computers and/or tablebases. My feeling is that chess is practically solved now .

That is not correct logic. It is more likely, and has been shown that all humans suck at chess, but some just suck less. Even when given time to put it in book format. As chess engines keep getting stronger, not humans. So chess is not "weakly solved". Until we see no more progress in chess engines. And that has not happened yet.

And remember his book is called. How to think like a grand master.

The title says it all in the light of the depth of chess.

The correct and more accurate title in today's understanding should be. How to suck less at chess.

[jp]

all humans suck at chess, but some just suck less. Even when given time to put it in book format. As chess engines keep getting stronger, not humans. So chess is not "weakly solved". Until we see no more progress in chess engines. And that has not happened yet.

Even when we see no more progress in chess, it does not mean chess is weakly solved, however we wish to define that. It just means computer chess has hit a plateau.

[jdart]

Human endgame analysis (pre-computer, pre-tablebases) is even more deeply flawed. I think it's fair to say that most of it won't stand up to scrutiny in the current computer era.

[Albert Silver]

This position was shown in page 20 of Think Like a Grandmaster by GM Kotov.
Don't get me wrong, I am a big fan of this book and of Kotov too. In fact I even read/ analysed this book twice during my teenage days.
The communication during those years when Kotov wrote the book was by snail mail (in postal service) and the variations was
then printed in chess journals or newspapers chess section.

Flohr-Fine 1935/6
[d]r1r3k1/p4ppp/bpq1Nn2/4R3/1PBR4/2Q1PP2/6PP/6K1 w - - 0 1
Here, Flohr made a losing move 24.Nd8? which made R. Fine win this game.
Quote from the book:
"Annotators the whole world over analysed this position. A win for White found in one country was quickly refuted in articles published in another. A practically invisible finesse spotted by one analyst was soon shown to be an error on further examination. Finally the English master Winter found the one and only way to win."

...one and only way to win... (24.b5!)

The funny thing is that SF+NNUE said 24.Nxg7! is even stronger than 24.b5! which points out that Kotov is wrong when he said "one and only way to win with 24.b5".
Ok, to be fair with GM Kotov 24.b5 is 2nd best line for SF-NNUE on my very old machine.

This is actually pretty common, and is not exclusive to this book, much less SF. I have found plenty of refutations from classic works through personal analysis that I confirmed with an engine. This includes tactical exercise books, to works on the middlegame, and more. If you work through the positions from the Encyclopedia of Combinations (great digital companion to iChess on Android), you will find such as well. Either the main move on occasion or the continuation. Obviously it detracts from comments where he might write "the only...", but above all it is the method and process that he teaches that is of value. The rigor and discipline in analysis when needed.