: ( 648aa)
White plays Dany's atekomi, intending to start Dany's Ko a few moves later.

Please note that this move is played outside the Main Semeai, but does not lose a tempo therein, because becomes an approach-move liberty of White's group at the left.

Black is still par with the Punishment Semeai Liberty Level.

Black is unable to continue with , or , in the left centre, because any of these moves would push him below the Punishment Semeai Liberty Level.
On the other hand, White threatens the severe cut at , which would win her the game.

Black is strictly bound to continue with the Main Semeai. Therefore, giving atari at is his only choice, because this move is the only one that also defends against White's cut at (which White will not play now; please refer to this subvariation 648of for details).

White will connect at , providing Black with two Main Semeai Liberties. Now, Black is able to defend against White's cut at .

We ask for your kind understanding that we felt unable to adjust the move numbers (i.e. to increase these by two) in the many already created diagrams, which follow in this part of the variation tree.

: (A 648mm)
Black intends to start Dany's Ko.

White will win this ko anyway, so Black can simplify the issue by retreating to , instead.

Black's atari is played outside the Main Semeai, and so loses a tempo therein.

:
White creates a ko shape.

Again, her move is played outside the Main Semeai, but creates another approach-move liberty for her left hand group at C. White does not lose a tempo.

(Referenced by 648cy)

: (A 648ms; B 648mz)
Black captures the ko in the centre.

As two moves earlier, he might retreat with , instead.
Defending his lower left corner with , however, is a grave mistake. Black does not have the time to do so.

Black's move is played outside the Main Semeai, and so loses a tempo, in principle. However, White's two approach-move liberties vanished with this move, so the net effect on the Main Semeai Liberties is plus one.

:
White captures the nakade as a ko threat.

Black is at nine Main Semeai Liberties now.

Winning the ko will require two White moves at (capture), and at (connect). Both of White's approach-move liberties will become effective again, so the net effect on the Main Semeai is zero.

Black, however, will lose a tempo in the Main Semeai by his connection at , should he be able to win the ko.

Please note that - from now on - we will not update the board's border colour after moves that capture the ko in the left centre, as long as this ko fight is still active, and undecided.

: (136 648na)
White is forced to connect at , following the usual course of the Main Semeai.

Before finding our earlier mistake with playing White's kikashi , in the centre too early, Dany questioned, whether it might be feasible to generate an additional White ko threat by splitting the Nakade Sequence here. But our deeper analysis showed that it is not!

:
Black connects, erasing his potential shortage of liberties in this area.

:
White plays her kikashi in the lower left corner.
This move loses one tempo in the Main Semeai, but gains one point overall, because Black will have to play an additional defensive move in the corner, after White won the ko in the centre.

: (A 648nc; 141 648nd)
White connects her single stone the upper edge, losing another move in the Main Semeai.
Black is four over par now. Please remember that the calculation of Main Semeai Liberties is based on the presumption that White will win the ko fight, and connect the ko.

Counter-intuitively, it will be worse for her to block at , instead, erasing several potential Black ko threats at the top.
Also, White must not connect the ko, so ending the ko fight (too) early. Black would benefit from capturing at unnecessarily.

:
Black recaptures the ko.

Ko: ; ;

; : (A 648qa)
Please understand these two moves as replacement for White's kikashi , in the centre, which had been played too early by us.

We now know that White's atari in the upper left centre cannot be counted as an additional ko threat.

While working on another topic, Joachim discovered by the way that White's push at – that has used by us for so long as a ko threat – does not include a sufficient threat, which would enforce an immediate local reply by Black.
The same is true for a White move at .

In the subvariation, we will use White's push at the left edge, instead, to explain Joachim's finding in detail.

Ko: ; ;

:
Black's ko threat loses one tempo in the Main Semeai. Black is back at three over par.

: (A 648ne)
White captures Black's single stone at the top.

It will not make a difference, if she captures at , instead. Thereafter, Black has the atari at as a ko threat.

Ko: ; ;

:
White's ko threat loses a tempo in the Main Semeai. Again, Black is at four over par.

:
Black's ko threat loses a tempo in the Main Semeai. Black is at three over par.

: (168 648nf)
Black is forced to answer in the upper left corner.

Connecting the ko in the centre, instead, will be a grave mistake.

:
Black's last ko threat loses a tempo in the Main Semeai again. Black is at two over par.

:
White's ko threat forces Black to occupy a point (), which he would have to take anyway during the course of occupying White's liberties at the left, so White loses a move in the Main Semeai here. Eventually, Black is three over par
We chose this move as White's final ko threat (instead of in the upper left corner) to explain its special properties in detail.

Compared to White's territory-reducing (in the Capture Variation) ko threat at (Black would answer at ), White's ko threat in the centre does not affect the score of the game here.
would provide White with a gain of one point in the Capture Variation (Black has one point of territory in the upper left corner less); while she would lose one point in the Semeai Variation (Black would capture an additional prisoner). This implies that - in comparison, from White's point of view - White's really played move "lost" one point in the Capture Variation, but "gained" one point in the Semeai Variation.
This effect is already known to us - it is typical for Black's territory-reducing moves, e.g. at . In principle, White provided Black with a territory-reducing move of his own.
However, Black will not benefit from this "donation". The number of territory-reducing moves that he really wants to play is strictly limited, due to their negative effect in the Semeai Variation. With the exchange of , played, Black will simply play one territory-reducing move less than without this exchange, but the final scores of the game will remain unaffected.

However, please be very well aware that the above is true, only if Black does not have a valid ko threat left (outside the Main Semeai area) that he could use with Joachim's Final Ko in the Semeai Variation ( 648ag).
If White's peep was played too early (e.g. before Black played his atari at at the top as a ko threat), Black would terminate the ko fight thereafter, in order to win the Semeai Variation by 15 points.

:
Black does not have any valid ko threat left, so he continues with the Main Semeai.

Joachim found that Black's throw-in at in the upper right were a valid ko threat.

White is barely ahead in ko threats (, vs. ), so our earlier mistake does not affect the outcome of the game. White will still win the ko fight, even with one effective ko threat less.

:
White has to connect the ko.

Otherwise, Black will recapture the ko soon, and connect thereafter, gaining an advantage of two points in the Semeai Variation.

:
Eventually, White plays her remaining kikashi in the lower left corner.

Black plays 1/3 reductions.

Before the end of the game, Black will have to capture at .

Capture Variation:

Compared to the superordinate variation ...

• White's cumulated gain at the upper edge is two points.
• Black did not get any territory in the left centre, so losing one point.
• Black destroyed four points of White's territory in the upper right, so gaining four points.
• Black got one point of territory less in the upper left corner.

In total, gains and losses compensate for each other.

White + 2 => White + 2

Semeai Variation:

Compared to the superordinate variation ...

• White's cumulated gain at the upper edge is two points.
• Black did not get any territory in the left centre, so losing one point.
• White captured an additional four Black stones in the upper right, so gaining four points.
• Black captured an additional White stone at the left edge, so gaining one point.

In total, White gained six points.

Black + 4 => White + 2