nocteus wrote on 12/28/18 at 11:07:24:
Hi, when I look at the KID's history i see only great attackers like Kasparov, Geller, Bronstein, Stein, Smirin, Radjabov,...
You forgot Tal! There were also some universal players: Fischer, Gligoric, and Najdorf for instance. Smyslov and Petrosian were more positional players and not known for the King's Indian, but they would play it sometimes. Larsen didn't like it at all but would still play it occasionally.
nocteus wrote on 12/28/18 at 11:07:24:
Can the KID be played strategically (à la Andersson for instance)?
What I mostly found are games (and books) where Black makes huge positional concessions (often giving up on a side) to obtain activity and unclear positions.
I think the popularity of these bridge-burning lines with books is because checkmate sells. And their popularity with masters is because the current engines on consumer hardware can still mis-evaluate them. A player of Kramnik's calibre is not going to follow the computer down the rabbit-hole, but lazy white preparers who lean on the spacebar too much are going to be punished. Of course to score in this way black also needs to be very well prepared. And it's risky. Kasparov had something interesting to say about why he gave up the King's Indian, you can read it here:
https://en.chessbase.com/post/three-draws-before-the-final-rest-day . I foresee the day when computers are just too strong, and the Mar del Plata variation will be written off -- not because it is so bad, but because it will be too easy for white to prepare, and the risk/reward becomes too unfavourable for black.
nocteus wrote on 12/28/18 at 11:07:24:
Can you play the KID as black in a more balanced way without necessarily burning all of your bridges, favoring for instance a game of manoeuvre and small nuances with long-term advantages (a bad piece vs a good piece for instance...) ?
Well, first of all, in the blocked oblique pawn structure (e4/d5 vs e5/d6) the bad piece is the one on g7, so that's worrying. On the other hand, the King's Indian is certainly sound and logical. It is also very useful as a winning try against some overly-cautious whites. Computers are showing more and more openings end up 0.00 with best play, and if you carefully examine some of the old "forgotten" lines there are probably many that could be rehabilitated, at least towards equality.
Just as one example, in
Dangerous Weapons they give this strange line:
1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.Be2 O-O 6.Nf3 e5 7.O-O Nbd7 8.Be3 Re8!? 9.d5 Nh5 9...Ng4 is the old move, e.g. Gligoric - Geller, Monaco 1967
http://www.chessgames.com/perl/chessgame?gid=1049259 , and see Barden, Hartston, and Keene (1973)
The King's Indian Defence, pg.230-231
10.g3 Bf8! (their exclamation mark) Gelfand - Radjabov, Corus 2008
http://www.chessgames.com/perl/chessgame?gid=1482382 Not 100% convincing, but we get the impression that there is more than one way to play the King's Indian.