Forums
Forums

Is the King's Indian Defence refuted by stockfish?

Sort:
asmitadeb

Stockfish doesn't seem to like the King's Indian Defence. It thinks white has too much of space advantage. So is it safe to say that the King's Indian Defence has been refuted?

TheNumberTwenty

No, not even close. KID is one of the best openings even at the grandmaster level for creating imbalances

TheNumberTwenty

Someone should tell you this early so you know... If an engine says an opening is +1 or -0.8 or something like that it doesn't mean an opening is "refuted" because you have to consider the depth level of the engine as well as human capabilities especially when you're employing the opening against amatuer players. Stockfish can destroy the King's gambit but when 2 humans play it both sides have good chances due to the double edged tactical nature of the position (which is irrelevant to engines since they won't miss a tactic pretty much ever)

yourverysister

not particularly

mejurist

I have seen this brought up over and over but have not seen the sufficient evidence. Engines are also notoriously untrustworthy when it comes to openings. You most of the time want a human position that you feel comfortable, not the marginally higher evaluated top engine line. 

tygxc

#1
Also AlphaZero prefers the Grünfeld Indian Defence over it.
Yes, King's Indian Defence is very risky.
Fischer did not dare to play it against Botvinnik, Taimanov, Petrosian, or Spassky.
'King's Indian Defence is riskier for black than King's Gambit for white' - Bronstein
Kasparov and Radjabov gave up on the King's Indian Defence.
This was Kasparov's last tournament game with the King's Indian Defence
https://www.chessgames.com/perl/chessgame?gid=1070932

ICCF GM Nefedov tried to rehabilitate the King's Indian Defence with 9...Ne8, but he suffered two losses in the 30th ICCF World Championship.

TheNumberTwenty

So yea, not currently fashionable at the world champion level, but I promise you it's fantastic against 99.99 percent of players. Good and easy system.

tygxc

#7
"Good and easy system"
++ Good to play for a win at higher risk of losing, yes.
Easy, no. One inaccuracy and you are lost.
#8
"can't share any lines"
See the Kramnik - Kasparov game above
Here are the 2 losses by Nefedov: black gets overpowered on the queen's side




TheNumberTwenty

I'd argue that it is easy, you have a very clear plan as black and the hard part is knowing when to apply the breaks and respond to whites threats on the queen side. Yes, it definitely isn't that good at the very top level, but I still think it's fantastic for most players who want to create imbalance with the black pieces.

tygxc

#10
"you have a very clear plan as black"
++ That is true, but you have to be painstakingly accurate in its execution: one small omission and you are lost.
King's Indian Defence was the favorite of Bronstein, Geller, Tal, Fischer, Kasparov, Radjabov.
It was great to play for a win with black even at higher risk of losing.
The point is that to win a tournament it is not enough to win one game with white and draw all your other games. To win a tournament you need to win against the weaker participants regardless of color. The King's Indian Defence was perfect for that.

idilis
asmitadeb wrote:

Stockfish doesn't seem to like the King's Indian Defence. It thinks white has too much of space advantage. So is it safe to say that the King's Indian Defence has been refuted?

Yes. We're all doomed. Next opening.

But before that...

https://www.chess.com/forum/view/game-analysis/why-do-the-chess-com-computers-always-overestimate-space-advantages

tygxc

#12
The problem is not the space advantage, space is overrated indeed.
The problem is that the white plan:
advance queen's side pawns, create a weak pawn, capture it, make a passed pawn, queen it
is more likely to succeed than the black plan:
advance king's side pawns, sacrifice, checkmate the white king

Vegosiux

I think people are putting too much faith in engines. Stockfish isn't the chess pope speaking ex cathedra nor is its analysis the holy gospel of chess gods.

There is only one way to "refute an opening", and that's playing against it and winning the game.

tygxc

#14
What else? 7...exd4
https://www.chessgames.com/perl/chessgame?gid=2069225 
...Nh5 was demolished in Kramnik - Kasparov above. Is there any improvement?
It caused Kasparov to give up on the King's Indian Defence.

blueemu
TheNumberTwenty wrote:

Someone should tell you this early so you know... If an engine says an opening is +1 or -0.8 or something like that it doesn't mean an opening is "refuted" because you have to consider the depth level of the engine as well as human capabilities especially when you're employing the opening against amatuer players. Stockfish can destroy the King's gambit but when 2 humans play it both sides have good chances due to the double edged tactical nature of the position (which is irrelevant to engines since they won't miss a tactic pretty much ever)

This.

Another point to bear in mind is that humans don't play chess the same way that computers do. The process of analysis and move selection is handled completely differently by a human or by an engine.

To an engine, there's no such thing as a position that's difficult to analyze or difficult to play. Excepting those positions that contain a clear forced win (mate or decisive loss of material), all positions are equally difficult (or equally easy) for an engine to analyze. It follows the same process, the same series of logical steps, in every case.

This is certainly not the case with humans. Humans can recognize a particular position as "complicated", or "unclear", or "wild". These categories make no sense to an engine.

tygxc

#17
"To an engine, there's no such thing as a position that's difficult to analyze or difficult to play."
++ There is. They are better at open positions than at closed positions.
They are bad at endgame fortresses.

blueemu
tygxc wrote:

#17
"To an engine, there's no such thing as a position that's difficult to analyze or difficult to play."
++ There is. They are better at open positions than at closed positions.
They are bad at endgame fortresses.

That's because they use the same, non-human, method of selecting moves regardless of the position they face.

You are talking about results. I am talking about methodology.

tygxc

#19
Yes, they use the same methodology regardless of the position,
but closed positions and endgame fortresses are more difficult to analyse/play correctly for them, i.e. require more time to  correctly evaluate.
Back on topic: the Bayonet Attack of the Mar del Plata Variation of the King's Indian Defence is an example of a closed position, where engines are weak at evaluating.

Optimissed
asmitadeb wrote:

Stockfish doesn't seem to like the King's Indian Defence. It thinks white has too much of space advantage. So is it safe to say that the King's Indian Defence has been refuted?

Stockfish is useless at positional assessments. It's best at endings. Openings are all positional, after all, so it can't assess them. I often play continuations in the late opening or early middlegame which it assesses as -0.4 and it's hopelessly wrong. They resolve into near equality with best play.

In a Chess960 game last year, when I was going through it, it was impossible to get a proper assessment out of the analysis tool they have here because it was obsessed with the idea that I should play either a4 or h4, can't remember which. It had that as its favourite move for about six moves and ruined its own analysis. Humans are better at positional analysis than computers.

Optimissed
DubstepJunkie wrote:
tygxc wrote:

#14
What else? 7...exd4
https://www.chessgames.com/perl/chessgame?gid=2069225 
...Nh5 was demolished in Kramnik - Kasparov above. Is there any improvement?
It caused Kasparov to give up on the King's Indian Defence.

7...Nbd7?

I'm thinking 9. ... Bd7!?. I suspect ...c6 is a mistake.