Hello everyone.
Below is an annotated game featuring the Samisch Variation of the King's Indian
Defence. There are various instructive points particularly at moves 23/24 and
moves 30/31. At the latter point things get very complicated so I would
strongly advise you to have two boards handy.
Enjoy and learn!
Tyson
Text begins:
(10) GIBSON,C - MORDUE,A T [E87]
4NCL, 07.12.2008
I occasionally get asked how do I play against the Samisch variation of the
King's Indian Defence. Here is one line which results in a tense positional
game for a while. Each side gets to execute a tactic that exposes the other
King but Black always has more pieces involved in his attack and he crashes
through in spectacular fashion.
1. d4 Nf6
2. c4 g6
3. Nc3 Bg7
4. e4 d6
5. f3
This is the signature move of the Samisch variation. The e4 Pawn is solidly
bolstered by the f3 Pawn. However, this denies White's King's Knight the f3
square and also weakens e3 and the e1–h4 diagonal. Nevertheless, it has a
fearsome reputation.
5... O-O
6. Be3 e5
7. d5
This closing of the centre denies Black any play based on ...e5xd4 and or
...Nb8-c6. It is regarded as the more aggressive of the two main alternatives,
the other being
7. Nge2. However 7. Bd3, developing this Bishop before the g1 Knight, is not a
good move because of
7... Ng4! 8. fxg4 exd4 ruining White's Pawn structure.
With the centre now closed each side must choose which side of the board they
want to operate on. A regular line for Black is 7... c6 and swapping on d5 to
open the c-file. In this game I adopted the older method of playing on the
Kingside.
7... Nh5
8. Qd2
If 8. g4 Nf4 Black is happy to sacrifice a Pawn on f4 to get an open e-file and
a1–h8 diagonal.
8... f5
There is a line involving an eccentric Queen sacrifice here which goes
8... Qh4+ 9. g3 Nxg3 10. Qf2 Nxf1 11. Qxh4 Nxe3 12. Ke2 Nxc4
with two Bishops and two Pawns for the Queen. This was pioneered by Bronstein
in his game with Spassky at the 1956 Candidates Tournament in Amsterdam. The
game was ultimately drawn However Karpov, against Velimirovic at Skopje 1976,
improved with 12 Ke2! and Velimirovic failed to gain any activity. Karpov went
on to win a long game.
9. O-O-O Nd7
In a Geller-Gligoric game from the 1953 Candidates tournament in Zurich,
Gligoric chose 9... f4. It seems odd to close the part of the board where Black
wants to play. Indeed, Gligoric was lucky to draw after Geller built up a
strong Queenside initiative.
10. Bd3 Ndf6
A sophisticated line that adds to the pressure on e4 but leaves the other
Knight somewhat out on a limb.
The tension between the e4 and f5 Pawns here is typical of this variation.
White would like Black to capture on e4 in a position where White gets to
recapture with a piece and use the square as an outpost. Black would like White
to capture on f5, recapture with a Pawn and then have the options of advancing
either ...e5-e4 or f5-f4 as appropriate.
There are various finesses such as White leaving his Bishop on f1 and Black
playing ...c7-c5 which enrich the position. Much to study here.
11. Nge2 a6
A flexible move that has two points. One is to play ...b7-b5 and open up lines
on the Queenside where White's King has taken up residence. However, three-time
Soviet champion Leonid Stein once played b7-b5 without preparation at all in a
wild game against Viktor Korchnoi in the 1965 Soviet Championship and won a
cracking game. Note: that if Black does play ...b7-b5 and White plays c4xb5
then after ...a6xb5 the c3 Knight may not be able to take on b5 because it is
defending both e4 and a2. This is one reason why White soon plays Kb1.
The other point of the text is to play ...Qd8-e8 and not be concerned with an
irritating Nc3-b5 threatening a fork on c7.
12. c5!?
Interesting but possibly premature. White normally waits to see what happens on
the Kingside before playing this break. Also if you take into the account the
variation with 7... c6 where Black opens the c-file, then there is the
difference that White is losing a tempo compared to that line.
12... Qe8
An equally intriguing reply. Black keeps the tension. He never plays ...d6xc5
in these sorts of positions because that gives White a clear initiative in the
centre. The capture ...f5xe4 is an option but, as explained earlier it is best
avoided if White can keep taking with pieces. However, an example of when it is
good to play it is the following line:
12... b5!? 13. c6?! - normally Black would not want to allow this wedge - 13...
b4 14. Na4 fxe4 15. fxe4 Ng4!
and White can't move his Bishop too easily without allowing either a fork on f2
or ...Bg7-h6 winning the Queen!
After 12... b5 White does better to play 13. cxb6 according to Deep Rybka.
One of the points of Black's move is to avoid the continuation 13. cxd6 cxd6
14. Na4!? with possible penetrations on b6 to worry about. Little features
again abide here. Black doesn't want his light-square Bishop traded for a White
Knight and, by the same token, White doesn't want his dark-square Bishop
exchanged for a Black Knight.
13. cxd6?!
It may be better to keep the tension by playing 13. Kb1 immediately. The King
doesn't want to stay on c1 because of possible tactics on the c1–h6 diagonal as
hinted in the last note, and also after opening the c-file a White Rook will
want to breathe fresh air from c1. Additionally the King protects the a2 Pawn.
In the game Black plans the advance ...b7-b5-b4 followed by ...Qe8-a4 tickling
said Pawn which is now defended.
13... cxd6
14. Kb1 b5
Now that Black has been allowed this advance with some ease the position is
roughly level. Nevertheless, it is still not entirely clear where both sides
aim to play. Black has a small initiative on the Queenside but White could play
down the c-file. Alternatively White could exchange Pawns on f5 and try to
advance g2-g4 with an attack whilst Black still has options of ...f5-f4 to
block White or ...f5xe4 to play in the centre.
This is a defining moment in the game. Black is threatening to simply play
15...b4 winning the Knight on c3 because it has no flight squares – 14. Kb1
took away the b1 square. White does not want to play 15 a3 because this will
make it easier for Black to open lines for an attack. What he does is create a
square on e2 by moving the other Knight but he first swaps Pawns on f5 so that
his 16. Ng3 threats the f5 Pawn.
15. exf5 gxf5
Deep Rybka claims Black is equal after 15...Bxf5 then changes its mind after
16. g4 to a small advantage for White. As already explained Black should
usually recapture with the Pawn here. However, the variation is worth
elaborating on. Deep Rybka has in mind 16... Bxd3+ 17. Qxd3 e4! - a typical
Pawn sacrifice in these lines – 18. Nxe4 Nxe4 19. fxe4 Nf6 - recovering the
Pawn because of the fork – 20. Nd4! Nxg4 21. Ne6 Nf2 22. Bxf2 Rxf2 23. Rd2 and
the two remaining minor pieces are powers to behold.
This goes back to my previous comment about small battles. The exchange of
Black's light-square Bishop leads to a weakening of e6 and we have a White
Knight proudly standing there. The swap of White's dark-square Bishop lets
Black corresponding piece dominate the long diagonal and an incursion on f2, a
key dark square, has also happened. Deep Rybka says after 23... Qf7 24. Rc2
White is better by 0.24 in this fascinating position.
After the text move the engine says that White is also better to the tune of
0.16. However, this is not about me choosing a move that a computer says is
better than the alternative, even if the difference is only a measly 0.08.
General principles say take with the g6 Pawn and this is because it gives Black
more options. The computer has spent time checking these lines. I just played
the recapture immediately and saved myself the bother.
16. Ng3
Consistent. The alternatives were 16. Ka1 to allow 16...b4 17 Nb1 but that's
just miserable, or 16 Bc2 to answer 16...b4 with 17 Na4. The latter makes sense
but Black could just play 16...a5 and continue advancing on the Queenside with
the a-Pawn instead. Now Black does not want to swap Knights because he will
merely open the h-file for White's Rook - it will developed effectively without
even having moved! -and he will lose control of f4 for what that is worth.
Factor in White's Pawn break of g3-g4 and he is doing well. So I continue to
play on the Queenside but White now has a flight square for his other Knight.
16... b4
17. Nce2
It should be noted that this retreat coupled with the swap on f5 has apparently
deprived the d5 Pawn of protection. However, it is obvious that 17... Nxd5??
18. Bc4 the errant steed is pinned and under threefold attack. Nevertheless,
this Pawn is one that both sides have their eye on. By the same token Black now
has to decide how he is going to defend his f5 Pawn.
17... f4!
It may seem hypocritical that after all the foregoing about the importance of
the e4 square that Black should give up control of it so easily. Yet, there is
an important qualification - White now can't get a Knight there easily.
There will be a swap on h5 - - but then the other remaining White Knight can't
go via c3 or g3 to get to e4 because both are controlled by Black Pawns. There
are routes via e2-c1–b3-d2-e4 and e2-c1–d3-f2-e4 but the first will be
prevented when Black plays ...a6-a5-a4 and even in the event of the second
canter Black already controls e4 with a Knight of his own!
18. Nxh5 Qxh5
Not 18... fxe3? 19. Nxf6+ and 20. Qxe3 and the intermediate check has cost
Black a Pawn. Recapturing with the Queen is more flexible. Her Majesty can come
back via e8 or f7 while the steed can only retreat to f6 so it may as well stay
there.
19. Bb6 a5
20 .Bc7!?
This move eventually results in a small crisis for Black.
Deep Rybka evaluates the position as almost level. I felt happy with my
position as I knew I had attacking chances on the Queenside. Deep Rybka
suggests here the startling 20... Re8 and after 21. Bxd6 Bb7 and it appears
that Black regains the Pawn quite easily. What I wanted to do was get my Queen
back to e8 and then play ...Rf8-f7-b7 and continue playing on the Queenside so
...Re8 is not in my thoughts. Okay, let's push the Bishop away.
20... Ne8
21. Bb6
Deep Rybka suggests two other moves; the odd 21. Qc2 which risks a pin on the
c-file, and the deceptive 21. Rc1. It follows the latter with 21... Qf7 22.
Bxa5!? - the a8 Rook is overloaded guarding a5 and c8 because 20... Ne8 cut off
the other Rook's link with c8 - 22... Bf5 23. Bxb4 - no time for 23. Bxf5
because 23... Qxf5 is check and then there's 24... Rxa5 winning a piece - 23...
Qxd5! - possible because 24. Bc4 is illegal because of the pin on the b1–h7
diagonal and 24. Bxh7+ is also illegal because there's a Bishop in the way!
Deep Rybka concludes that it's a draw after 24. Bxf5 Qxa2+ 25. Kc2 Qa4+ 26. Kb1
Qa2+ 27. Kc2 Qa4+ etc. It's evaluation after 25 (27)... Rxf5 is dead level!
So to save all this calculation one plays 21. Bb6 immediately and presses the
clock...
21... Nf6
And now that we've pushed the pesky Bishop away we'll get on with the game.
22. Bc7
Oh! Are you back again?
22... Ne8
23. Bb6
Now I'm not interested in a draw. I can see a way to play on but it has a
drawback. Sometimes playing on involves risk so I'm just going to have to take
a chance that he doesn't see what is wrong with my move.
23... Rf7!?
24. g4?!
This is not a bad move, indeed it's the engine's second choice. However, White
misses his opportunity to get his Knight to e4 with the finesse 24. Ng3! Black
can't capture because 24... fxg3? 25. hxg3 traps the Queen.
23...Rf7 took out a flight square and the White Queen suddenly has relevant
control of g5.
After the text 24... fxg3 can be answered either by 25. hxg3 Qxf3 26. Bxh7+ and
White has a small advantage, while Black has the more uncomfortable King, or
25. Nxg3 with a bigger plus apparently. This is partly based on the line 25...
Qxf3? 26. Rdf1 Qg4 27. Bxh7+ Kf8 - 27... Kxh7 28 Rxf7 nets an Exchange – 28.
Rxf7+ Kxf7 29. Rf1+ and White is better than the previous line.
Of course Black has no interest in opening Kingside lines for White's benefit.
24... Qh4
25. Ng1!?
Although this move meets with a measure of success it seems simpler to play 25.
Rc1 with the inconvenient threat of 26 Bxa5, once again taking advantage of the
overloaded Rook on a8. Deep Rybka suggests 25... Rb7 26. Bxa5 Bxg4 27. fxg4
Rxa5 28. Rc8 Qe7 29. Bf5 with a mess that seems a slight plus for White. Note
that those key Bishops have gone again.
25... Rb7
26. Bf2 Qe7
27. Re1?!
This is clearly a prevention of an ...e5-e4 break. Black would be willing to
give up a Pawn to unleash his King's Bishop and there are lines such as
...e5-e4, Bxe4 Bxg4, fxg4 Qxe4+ picking up the h1 Rook because the Knight on g1
interferes with the Rook link. However, White doesn't have to take on g4 and I
think the g-file has been opened to his benefit anyway.
It seems more natural to play the immediate 27 h4. Deep Rybka advocates 27. Ne2
... Hmm. Of course if White is really concerned about ...e5-e4 then he could
just play
27. Be4.
27... Qf7
This is a natural enough move which is endorsed by the engine. It steps off the
e-file and avoids any tactics based on Qxf4 should the Black queen be
undefended for any reason. What I didn't realise was that I was walking into a
different tactic and needed to take some care.
28. h4 Nc7!?
Now Deep Rybka claims that this is a mistake but this depends on the judgement
of the position at move 31 and maybe well beyond! My reason for playing it was
that it gave me the option of ...Nb5 and I probably would have played it if
White had gone 29. Be4.
The engine claims that 28... b3, 28... a4, and 28... Nf6 are all better. I
wouldn't play the first because 29 a3 blocks the Queenside. The second would
make sense if I was planning ...a5-a4-a3 anticipating the reply b2-b3 and
answering that with ...Nc3+, and for that I need the Knight to have access to
c3. Even so it doesn't do that much because White's 27. Re1 at least had the
benefit of moving it out of a potential fork.
The third one, 28...Nf6, has a side benefit that I didn't appreciate, even if I
had actually considered the move! Now I normally would be reluctant to give an
oncoming Pawn Roller an extra tempo but here it doesn't seem to matter too
much. If 29. g5 Nxd5 30. Be4 Nc3+! 31. bxc3 (if 31. Ka1 Qxa2 mate) 31... bxc3+
32. Bxb7 Qxb7+! wins the White Queen with no problem. If instead 29. g5 Nxd5
30. Bc4 Nc3+! 31. bxc3
(or 31. Qxc3 bxc3 32. Bxf7+ Kxf7 leaves Black a pawn up with a huge position.)
31... bxc3+ 32. Ka1 Qxc4 is good, but there is also 32... cxd2 33. Bxf7+ Kxf7
and Black threatens 34... dxe1 and 34... e4+!
If 35. Re4 Black just promotes on d1.
Also 29. g5 Nxd5 30. Bc4 Nc3+! 31. White King moves are met simply by 31...
Qxc4.
Ironically after 28... Nf6 guarding d5 with 29. Be4 no longer works. After
29... Nxe4 30. fxe4 drops the g4 Pawn while 30. Rxe4 Rb5 wins the d5 Pawn!
Deep Rybka says that after 28... Nf6 White's best is 29. Nh3, so after 29...
Nxd5 we would transpose to the game.
29. Nh3
Now Deep Rybka says I should play 29... h6. Of course one doesn't advance Pawns
in front of one's King in the face of an oncoming Pawn storm unless necessary.
Anyway I think that I have a better move, with threats.
29... Nxd5
And now the fun really starts.
30. Bxh7+!?
This came as a surprise. I'd half-expected 30. Ng5 but then 30... Nc3+! I'd
calculated:
a) 31. Ka1?? Qxa2 mate;
b) 31. bxc3 bxc3+ 32. Ka1
(or 32. Kc2 Qxa2+ 33. Kxc3 Qb3 mate)
32... cxd2 33. Nxf7 e4+ mating in two more moves;
c) 31. Qxc3 bxc3 32. Bxh7+ Kh8 33. Nxf7+ Kxh7 and White is losing because the
f7 Knight is en prise and Black also threatens 34...Rxb2+ winning the Bishop on
f2.
Other moves are:
d) 31. Kc2 b3+ and now
d1) 32. axb3
(32 Kxc3 Rc7+ 33. Bc4 Qxc4 mate)
32... Qxb3+ 33. Kc1 Na2+ 34. Kb1 e4 with a winning attack now that the Bishop
joins in the siege of b2. Or
d2) 32. Kc1 bxa2 33. Bxh7+ Kf8 34. Qxd6+ Qe7 35. Qxe7+Rxe7 36. Kc2 Nb1+ forces
White to give up an Exchange with 37. Rxb1 to prevent a free promotion. I
suppose here Black can add more humour to this crazy game by underpromoting to
a Knight with check!
e) 31. Kc1 and now:
e1) 31... Qxa2 32. Bxh7+ Kh8 33. bxc3 Qa1+ 34. Kc2 Qa4+ 35. Kc1 Qa1+ 36. Kc2
Qa4+ with a draw. Or,
e2) 31... Nxa2+ 32. Kd1
(32. Kb1 Nc3+ 33. Kc1 Na2+ is yet another drawing sequence which I'd spotted so
I knew I probably wasn't losing.)
32... Qb3+ 33. Bc2 Be6!? 34. Nxe6
(34. Bxb3 Bxb3+ 35. Ke2 Bc4+ 36. Kd1 Bb3+ with another draw!)
34... Qxf3+ 35. Qe2 Qxd5+ 36. Nd4 exd4 37. Qe4 and after all this nonsense this
Deep Rybka still says it is equal!
The other move that I anticipated was after 29... Nxd5 was 30. Ka1 to avoid
30... Nc3+ with check.
Regardless I intended 30... Nc3 31. bxc3 e4 which I thought was winning.
However, Deep Rybka went further and found 32. Rxe4! Bxc3+ 33. Qxc3 bxc3 34.
Ng5! and claims White is better. There are suddenly a lot of pieces pointed at
the Black King. Unfortunately there is the simple line
30. Ka1 Nc3 31. Bxh7+ Kh8 32. Bb1! defending the mate. Deep Rybka says Black
should play after 30. Ka1 the equally simple 30... h6.
So what about the text move 30 Bxh7+ then? After I bit of reflection I thought
it was a mistake for several reasons. Firstly the move has been executed so is
no longer an option in a different line where it may be more important.
Secondly although I can't take it immediately because of the obvious fork the
Bishop remains en prise. If my Queen moves somewhere, for example taking on a2
after the moves ....Nc3+, Kc1 then the fork is no longer viable. Also if I play
...e5-e4 and the Bishop moves down the diagonal then ...Qxh7 is a move. Finally
with the h7 Pawn gone further White Pawn advances don't open lines if Black
keeps moving things out of the way. Anyway, Black's reply is obvious and the
ball is back in White's court.
30... Kh8
White's simplest move here is probably 31. Ka1. White's problem is that 31...
Nc3 is currently coming with check followed by an incursion with Queen or
Knight on a2, and that he can't take bxc3 because ...bxc3 in reply is check and
it also attacks White's Queen.
Deep Rybka does also advocate 31. Qc2 but suggests two decent replies;
a) 31... b3 32. axb3 Nb4 and White's King position is compromised and Black has
a fine outpost on b4. I think it underestimates Black's position here.
b) 31... Nc3+ anyway. After 32. bxc3 bxc3+:
b1) 33. Ka1? Rb2 34. Qa4 Bd7! - pushing the Queen away from control of d4 – 35.
Qa3 e4! wins as the Bishop is uncovered;
b2) 33. Kc1 Rab8 34. Ng5 Qc4 and Black has a huge initiative. His simple threat
is 35... Rb2 followed by 36... Qxa2 and 37... Qa1 mate.
The engine flounders with its assessment going from White pluses to Black ones,
but the key factors are the badly placed White King and the White Kingside
pieces that don't come to the defence quickly enough.
Back to 31. Ka1. I mentioned in a previous note that if 31... Nc3 White could
simply play 32. Bb1. After some reflection Black may be able to answer 32...
Nxb1 33. Kxb1 or Rxb1 d5. with a sensible position. He has the two Bishops in
an open position and a good Pawn centre.
Originally after 31. Ka1 Nc3 Deep Rybka wanted to take on c3 claiming White
stood much better. Its analysis went 32. bxc3 bxc3 33. Qxc3 Rc7 34. Qd2
(if 34. Qb2 Rb8! 35. Qxb8 e4+ wins)
34... e4+. At this point the engine changed it's opinion radically. Analysis
went on
35. Bd4 Rc2!
(not that 34... e4 has interfered with the h7 Bishop's control of c2)
36. Qxc2
(if 36. Bxg7+ Qxg7+ mates)
36... Bxd4+ 37. Kb1 Rb8+ 38. Kc1 at which point White's King is now running in
the opposite direction to the one it started this line with!
The engine now claims that after
38... Qf6 Black is plus 1 or so. There are many variations but I shall just
give one:
39. Kd1 Be3 40. Rxe3 Qa1+ 41. Ke2 Ba6+
(note that White has two Rooks en prise)
42. Kf2 fxe3+ 43. Kg2 exf3+ 44. Kh2 Qe5+!
This line highlights that the White pieces lined up haphazardly on the h-file
don't contribute to the defence. The computer counts the material but not
Black's options. Look how late Black's Queen's Rook and Bishop are coming to
the party (moves 37 and 41) but they still have a lot of firepower.
Clearly there is a lot to analyse here but Black is not without practical
chances whether he has a material deficit or not. Gibson chose an alternative
defence but it fell to a simple yet serious oversight.
31. Rc1? Nc3+
32. Rxc3
After 32. Kc2 Deep Rybka says it is mate in ten after
32... b3+. Whatever the lines are Black is looking forward to playing ...e5-e4
and releasing his Bishop to great effect.
32... bxc3
33. Qxc3 Rab8
If here 33... e4 then 34 Bd4 defends. However, the text threatens b2 a
different way and White's own Bishop on f2 blocks the defence 34. Rh2.
34. b3
Afterwards Gibson admitted that he had failed to see that the Black Queen on f7
controlled b3. It makes no difference. If
34. Be4 Rxb2+ 35. Qxb2 Rxb2+ 36. Kxb2 d5 and 37... e4+,
or 34. Rd1 Rxb2+ 35. Qxb2 Rxb2+ 36. Kxb2 Qb7+ and 37... Kxh7, a scenario hinted
at earlier.
Or lastly 34. Bc2 Rxb2+ 35. Qxb2 Rxb2+ 36. Kxb2 Qb7+ and 37... Qxf3 among other
things.
34... Rxb3+
35. axb3 Rxb3+
36. Kc2
Whatever White plays Black is going to shortly have a Queen check to avoid the
Knight fork on g5 and then he will take the h7 Bishop leaving White with only a
Queen for the Rook. As it goes Black may be mating the White King as it
advances into the desert that used to be his Queenside.
36... Rxc3+
37. Kxc3 e4+
Black's dream move in the King's Indian Defence. The engine says it is mate in
6 or 7 moves but there is no point in playing on.
38. 0-1
Text ends.
