Tips against king's indian defense

https://www.chessgames.com/perl/chess.pl?yearcomp=exactly&year=&playercomp=white&pid=21136&player=&pid2=19233&player2=&movescomp=exactly&moves=&opening=E80-E89&eco=&result=

Spassky - Fischer

Sämisch Variation

Spassky won 1, drew 2, lost 2.

Play the Tromposky Attack 2. Bg5 if 2.... g6 B,x g6 and black equalized.

The Bayonet attack can only occur if Black plays Nc6, giving White an instant free tempo. The idea is that after b4 a5 bxa5 Rxa5, Black cannot stop White from pushing his a-pawn forever, since the Rook cannot maintain its position, and White gains space on the Queenside. There are other ways for Black to play rather than ...Nc6 here. And none of these lines are refuted...Black can pretty much hold a draw in virtually any variation where White has less than a pawn advantage, with perfect play, but this is definitely the most dangerous system.

The f3/Be3/Qd2 systems, or f3/Be3Nge2 (without Qd2) systems take some of the pressure off Black if he is well prepared.

Then there's the 5 h3 systems, and then the 5 Bd3 0-0 6 Nge2 systems, which would seem to give Black an easy game, after 6 ...Nc6 7 0-0 e5 8 d5 Nd4, but White has several options to force accurate play by Black, including 9 Bg5, 9 a4 ((a bit of a positional trap) - ...a5 10 Nb5!? or 10 Nxd4 exd4 11 Nb5), or 9...c5 10 Nb5 Ne8 11 f3 f5 12 exf5 gxf5 13 Ng3!? f4?! (not best) 14 Ne4 Bf5 15 g4 or Bd2 or the tricky 15 b3, baiting Black into 15....a6?! 16 Nxd4 cxd4 17 g3! Nf6 18 Nc5! Bh3 19 Ne6 with an obvious Bishop Pair advantage (Look at that horrible piece on g7!)

At the highest levels, the King's Indian simply puts too much pressure on Black if White has good technique and understands all of the nuances of both queenside and kingside play and structures. The Grunfeld is a much stronger option (which was played at the recent super GM Tata Steel tournament--that should tell you how reliable it is).

However at the recent Tata Steel Tournament the KID was played 8 times while the Grunfeld was only played 4 times.

https://en.wikipedia.org/wiki/Tata_Steel_Chess_Tournament_2024

The KID has had a question mark over it when playing against the absolute best players for many years due to the Bayonet Attack and some other schemes but so has the Grunfeld. Between the two black defences going for a win the Grunfeld appeared to have won out for a number of years as the super gm opening of choice when black was being aggressive for a win, however this Tata Steel proves the KID is back again.

I should have paid more attention to the games. Not going to lie, I didn't really pay attention to much except a few games.

@25

"this Tata Steel proves the KID is back again"
++ Not really only sidelines and played by outsiders
E60 Donchenko - Wei 0 - 1 White plays e3 instead of e4
E60 Giri - Wei 1/2 - 1/2 White plays e3 instead of e4
E94 Ding - Donchenko 1/2 - 1/2 Black plays exd4 instead of Nc6
E94 Warmerdam - Donchenko 1 - 0 Black plays exd4 instead of Nc6

It doesnt refute it. It just makes blacks life harder but black can definitely get counterplay without having to deal with a forced loss.

Id recommend not allowing black to play f5 any time soon with an early d5 before castling

white hasnt commited to castling yet, so black expanding on the kingside can actually be punished.

@28

How?
Here are 2 more failed attempts:
https://www.iccf.com/game?id=948179

https://www.iccf.com/game?id=948250

In the 2024 Tata Steel Donchenko and Giri opted for a timid sideline as white against Wei.
Donchenko played the sideline exd4 instead of Nc6 to a loss and a draw.

@29
7 d5 makes it easy for black.
It is black who has to find a way not to lose against the Bayonet Attack.

black played sub optimally with 9.Ne8 10.Nd2 instead of 9.a5 or 9.Nh5.

9.Ne8 allowed white to transpose into an improved 9.Nd2 line (9.Nd2 a5!), and black often has a very hard time when white gets c5, Nc4, a4 Ba3, and a crushing queenside attack

To be fair, those games are from 7 years ago - before AlphaZero and resulting NNUE engines even came along.

I'm guessing improvements for Black have been found since then.

The KID is like the Sicilian in a way - certain lines rise to the top, and seem promising for a while. But modern engines have a knack for defanging even the most dangerous-looking variations ...

The chess.com game review says Kramnik played a much more accurate overall game than Kasparov, not just a better opening.

I put this "KID has been refuted" argument to the test by taking the Bayonet Attack, Sokolov line position and letting Stockfish 16 play it out against another Stockfish 16. The game ends in a draw.

Also it is not an unplayable line for black at master level. When black plays the best move 14...Nh5, there have been 10 games played, 3 wins for white, 5 draws and 2 black wins. Pretty standard kind of win rates for any opening.

@31

"black played sub optimally with 9.Ne8 10.Nd2 instead of 9.a5 or 9.Nh5."
++ Nefedov is an ICCF grandmaster and was playing in the World Championship Finals.
He surely knew why he tried 9...Ne8 instead of 9...a5 or 9...Nh5. 9...a5 10 bxa5 or 9...Nh5 11 c5 are problematic for black.

"black often has a very hard time when white gets c5, Nc4, a4 Ba3, and a crushing queenside attack" ++ That is the main problem of black in the Bayonet Attack. Black sets an ambitious goal to checkmate the white king and white sets a modest goal to weaken and win a pawn and queen it. If both play well, then black cannot achieve his ambitious goal, while white succeeds in realising his more modest, more realistic plan.

@32

"those games are from 7 years ago" ++ Started June 20, 2017, finished October 2, 2019.

"I'm guessing improvements for Black have been found since then."
++ I am guessing not: the King's Indian Defense has not been played in subsequent ICCF World Championships Finals.
In the Tata Steel 2024 Donchenko avoided the Bayonet Attack by giving up the center with 7...exd4 instead of playing the critical 7...Nc6.

@33
"The chess.com game review says Kramnik played a much more accurate overall game than Kasparov, not just a better opening."
++ But where did Kasparov err? What better move could he have played that did not lose?

"letting Stockfish 16 play it out against another Stockfish 16"
++ At what time per move? The ICCF games are at 5 days / move with engines.
Those closed positions are very hard for an engine because of the horizon effect.
A slight weakening on black's queen's side may cause the loss of a pawn 15 moves later and a pawn promotion 30 moves later.

"Also it is not an unplayable line for black at master level."
++At IM level, or at GM level in shorter time controls, or as a surprise it is playable, but at top level (candidates' tournament, world championship match, ICCF World Championship) it is out of fashion for good reason. Kasparov gave up on the King's Indian Defense after his loss to Kramnik and switched to the Queen's Gambit Accepted. Also Radjabov then abandoned the King's Indian Defense.

"When black plays the best move 14...Nh5, there have been 10 games played, 3 wins for white, 5 draws and 2 black wins." ++ Those numbers tell nothing. You have to look at the games. Stockfish evaluates 14...Nh5 at +0.9 that is a sure win.

I don't know the final verdict of the Bayonet Attack - but I'm betting it's a dead-draw if Black plays accurately - like most mainlines are.

9 ... a5 seems to be Black's simplest path forward. It's what Stockfish prefers. It's also what Carlsen plays.

Of course ... there's so much more game to play from there, and room for either player to err.

@35

I think it loses by force for black. I agree 9...a5 may be best. Maybe black can hold the draw.
Here is a 2021 correspondence game:

@38
"We don't to play like to play d5" ++ Fischer and Kasparov played both King's Indian Defense and Grünfeld Indian Defense, but trusting neither.

"white fianchettoing" ++ That is good for black. Kasparov had no trouble with Karpov fianchettoing as white in World Championship matches

"We like closed positions with a king ide attack and super agressive decisive play"
++ 'The problem is that although the goal of the kingside attack - mate to the enemy's king -
is more enticing, it is also far more difficult to achieve than, for example,
the creation of a few pawn weaknesses by means of a queenside attack.' - Bronstein

"The Saemisch (f3) " ++ That is good for black.

@34

after 9.Nh5 10.c5 Nf4 11. Bxf4 exf4 12.Rc1 h6 13.Nd4 g5 black is fine.

after a5 10.bxa5 Rxa511.a4 c5 12.Ra3 Ra6 black is still fine

black rarely gets a kingside attack. Doing it in the petrosian variation is unsound. Doing it in the averbakh is just dumb and impossible. Doing it in the saemisch is possible but black does it for different reasons and still has to find counterplay on the queenside