the story behind the game is interesting too. Piket had been winning game after game in this variation with a plan of transferring his bishop to the c8-h3 diagonal. Kasparov's 17th is his improvement on 17...h5 which had been played up until then.

John Nunn noted that the move activates the rook, leaves h5 free for a knight and that 21...Nh5 justified black's play. ("The Classical King's Indian")

The story gives an example of an important theoretical novelty from Kasparov. In 1988-89 some people were being put off a bit by Piket's wins in this line. I suppose one could argue that Jeroen was playing what he knew worked but was under no illusions of a forthcoming improvement. (As an aside, I guess the moral is also to avoid playing the same lines all the time or following something which seems to give an advantage in variations a to z as you never know if there is an improvement, the latter point once made by Bronstein.)

About 95% I would say and by that measure alone I don't agree it can be considered one of <Kasparov>'s best game. The same goes for Kasparov vs Anand, 1995 and Kasparov vs Anand, 1991.

<And even though piket had been winning game after game in this variation,you'd have thought he'd of known playing "arguably" the greatest ever player with over a 100 point rating gap,he'd have known Kasparov was far from likely to follow the path of his previous opponents in this variation?>

Well, this was played before computer DBs became de rigueur, when even the top players had narrower repertoires but, nonetheless, against <Kasparov>'s preparation, the expression that comes to mind is <You can run, but you can't hide!>

they've been playing 15. cxd6 for decades -- it's well established theory.