Is the King's Indian Defense bad?

If I remember correctly, even Nakamura seems to consider it to be a losing try at the top level.

I don't consider it a losing try, but it's having some problems at the top levels with the "Classical Variation." Last I heard, they're turning to the move 7...Nbd7, rather than 7...Nc6, since that over-used nonsense isn't working so well anymore. Also, the Exchange Variation, though still drawish, isn't as bad as previously thought.

When your opponent is near 2800 and spends months preparing with an engine, there's not much chance you can play it for a win, but at the same time it makes no sense to play it for a draw because there are much easier ways to do that.

So they don't play it. That's my guess.

Doesn't mean much for our level, although I will say white has many sensible tries that lead to very different positions, so as black you have to do a lot more studying. At our level that will matter.

Even at the lower levels, some players, like myself, are sick and tired of the 7...Nc6 shenanigans, and my issue is with the 13.Rc1 line.

However, I don't play 7...Nbd7.  I would play 7...Na6 before I played 7...Nbd7, but I've groan to love the 6...Qe8 line, which is gaining some popularity.  Maybe not at the 2800 level, but 2400 or so.

until you gain at least another 1000 ratings points and routinely compete with 2700's I would nit lose too much sleep over it.

Eh, Saemisch is fun to go up again.  The Fianchetto is more annoying.  Both games below I had a blast blowing away the Saemisch!

http://charlottechesscenter.blogspot.com/2019/06/points-schmoints-concept-of-compensation.html

http://charlottechesscenter.blogspot.com/2019/04/errors-in-opening-part-3.html

Yeah, stupid typo.  If only the A weren't so damn close to the W!  LOL!

Guess what,  Carlsen doesn’t even care about openings  and he is the no.1 

Well, for me, the King's Indian lives on.

For the rest of you fools ...

https://www.youtube.com/watch?v=9jK-NcRmVcw

As someone who plays the King's Indian, how do you deal with analysis when the computer gives evaluations that are wrong? For example, Nakamura blasted So one time for relying on the computer evaluation of these positions. http://chess-news.ru/en/node/19892

If the computers are wrong, what is the true evaluation of these King's Indian positions?

In an interview, in regards to a game against Nakamura, Carlsen says, "If he had wanted a worse position, he would have played the King’s Indian." Does this mean that Carlsen believes that the King's Indian Defense is bad? https://www.chess.com/news/view/carlsen-wins-2017-chess-com-isle-of-man-international

I have said this in numerous threads.  Computers are utter cr*p in the opening and the endgame.  If there are tablebases installed, they are reliable with 6 pieces or less, but 12 piece endgames, like King, Rook, and 4 pawns each, are not reliable.  It will tell you a side is better because of material count when it's not true.

I don't give evaluations based on a number.  I don't believe in the "7...Nc6 is -0.68, but 7...Na6 is -0.64, and so 7...Na6 must be better".  NO!  UTTER CR*P!  I see no purpose in giving a number.  My belief is that the King's Indian is just like any other "Normal" opening.  White is ever so slightly better because he goes first.  The same goes for the Nimzo-Indian, Queen's Gambit lines, etc.

The position that Nakamura is referring to in the article is on move 20. Do you consider move 20 to still be in the opening?

And what about other openings like modern benoni, benko gambit, and czech benoni? When the computer is giving +1 or so immediately, is it really wrong? The computer on your cell phone can beat the world champion at chess. And on the topic of endgames, if computers were weak at them, we would expect to see more humans victorious against computers by following the strategy of getting to an endgame and winning there. But this does not seem to be the case.

Yes.

In fact if you walk it down human lines, after many moves the +1 suddenly turns to +0.3
The same thing can happen when you walk it down it's own lines (actually put the engine recommended moves on the board and see what it thinks after 10 moves).

But this is a slow process, because after every move you put on the board you have to let it think again, and most weak players don't realize letting their phone think for 10 seconds is basically worthless so they never do this.

Other times it overestimates the structure and will perpetually give white +0.5 almost no matter what, over and over, and all the way into a dozen different 100% dead drawn endgames. I remember Grichuk commenting on a game that if the engine says 0.00 in _____ structure then actually white is in trouble, because the engine always overestimates this for white.

Engines play a lot of small mistakes, but it only takes one big mistake to lose the game.

Humans almost always make one big (at least big relative to the engine) mistake, so humans can't win... but still, engines make lots of little mistakes, which is why ICCF is still around. People who mindlessly select the first engine move without any further investigation are the lowest rated in ICCF.

hay

hi

Do you remember what structure Grischuk was talking about?

Then what do the highest rated players on ICCF say in regards to how they are successful?

Structures like this, with a pawn on d5, the engine will tend to say white is better for dozens and dozens of moves... right into a dead drawn endgame.

If you go back and have it try a different line, trying to let it prove itself right, it will lead to another drawn endgame, and another, and another.

Certainly ICCF players use engines to their fullest. They user proper hardware and software, let the engines (plural) think a very long time, etc.

But they will, for example, discard moves a human understands to be bad in the long term, and this allows the engines spend more time analyzing proper moves.

Engines are practically perfect in the short term, and they're getting better and better at long term considerations, but they can still sometimes be clueless about long term ideas.

Pfren plays correspondence on servers that allow engines. Sometimes he posts a game that he won because his opponent followed the engine too much.