I had not really calculated the score but ever since the middle game I had the idea that if I saved my group running from the upper left, I was ahead, probably by komi. So at move 126, seeing this move, which creates more eye space for my group while reducing that of the Black bottom right corner, I though this was enough.

It turned out it was, but by a margin of only 3.5

Also, there's a more urgent move at A, which destroys territory in sente and reinforces White's shape.

Anyhow we're in the early endgame now, where groups are safe but can still be threatened and where territories are sketched out but not all have clear borders yet.

Big moves are now (preventing)

• The aforementioned move at A
• Black's minor boundary play at B, which threatens to capture 2 stones
• White's jump at C, which reduces Black's territory while improving the health of the top group
• the 2nd line diagonal at D, reducing both territories and affecting their health
• White's 2nd line hane around E
• Black's hane at F
• Black's capture at G, which makes his group totally alive and weakens the white border stones
• the diagonal at H, affecting both groups' health
• White's jump down at I, providing eye space and threatening to connect to the corner

Move analysis

Move A

is sente for Black. If White doesn't respond, then Black A then B cuts off White's eyeless group.

is also sente. If Black doesn't respond, another move at threatens to cut the black group into pieces and will anyway cut off the b6 stones at the bottom.

With and , White makes a half eye for her group while reducing Black's territory. is sente because next either the 6 stones get cut off or Black's eyespace is reduced to one eye.

Hence Move A is double sente and has value for both, so it's the biggest move on the board.

Move B

here is small. It threatens to cut off two white stones, but while Black does that, White first executes the sente on the left side, then takes the big moves at and . White now leads by about 15 points, an effect of accumulated losses at (-7) and (-4).

here creates a half eye in gote. That is, White can now make one eye there but Black can still destroy the eye at .

Obviously Black has much better things to do.

Move B is a mere boundary play, affecting only two stones. It's gote for both and doesn't add much value. It's a late endgame play.

Move C

makes central teritory and also removes an option for White to create extra eye space in there, increasing the value of D (which White is probably going to take for herself).

does the opposite (obviously), creating more health in the center and reducing Black's potential to connect to the center. It increases the likelihood of Black playing D.

Overall, C feels like a big move in itself but it's not severe towards either opponent and it increases the value for the opponent to take D. So let's now discuss D

Move D

If White responds to with the marked here, then Black has made a gain in sente. So, WHite will rather respond at in sente (threatening to connect) and then make an eye at . We again see the relationship of these endgame moves.

Likewise , if answered at the marked point, is a gain in sente, so Black

Move E

This is a standard endgame move. After , Black will have a sente sequence shown by the marked points.

Similarly, after White has a half sente sequence resulting in the marked stones being the border.

Move E is a classical gote move with miai value 5, a large late endgame move.

Move F

Black can execute this forcing sequence first. Later the cut or connect at the marked point is big.

fixes White's troublesome shape and next can reduce Black's territory considerably. A turn at the marked point would affect all of Black's strings in the vicinity.

It's hard to assess F because it's a middle game like move which affects the power of many stones in a central area. Black's move there is locally forcing and forestalls White's reinforcement which affects the stones in the center.

Move G

After Black has a sente endgame sequence as marked. In this stage the territory is still open from the center and White's stones are weakened so he will leave the aji.

It might look as if is sente against Black's group here but this isn't true, as - shows.

Even if Black treats as sente, he can respond at and increase the pressure on White.

Move G is a big endgame move but not urgent at this stage.

Move H

If goes here without any preparation, White will respond with the big sente move creating a half eye, then play away because I and J are both available.

Likewise is answered by the sente on the left and then ignores the bottom because as shown G and K are miai.

In summary, move H is accomplishing very little at the moment for either side.

Move I

Since is undermining White's eyespace, it only makes sense to play it after the sente at A. Also, as a response (likewise after playing A in sente) connects and increases the value of H.

When White plays she has a large follow up at so locally the move is sente. After that, the value of H increases since White has invested in this area at the expense of the lower left corner.

Overall, I bears a relationship with the lower left corner.

Katago's order of play

White having a small lead, KataGo tends to play safely for White, notably at the top. After the marked exchange, Black takes the big move . As mentioned White can resist by answering in the center, but as I played it out, it is unclear whether White can survive without losing the marked stones. This would reverse the lead. Hence obliges and then forces again before playing out the sente in the center . With White's right side being reduced but strengthened, is the right timing. White would still win the game as per my playouts of the endgame but only by 2 points.

Concepts

Double sente

A first concept that arises from this analysis is double sente. A in this position is sente for both and with Black to play, it's a must play for Black. The question now is when this move arose to be double sente.

It turns out - obviously - that both had neglected this area in the fighting before and there had been an urgency from much earlier on for White to make an eye here while threatening to cut Black into pieces, while Black could destroy that eye in sente. Notably after White made a hanging connection at and Black defended his group at , the move at A was bigger than the pushing battle at the squared stones.

However, mistakes in middle game fighting happen and so, when groups are settled we should look out for such opportunities left behind by both. These are potential "double sente"

Central shape moves

The hane at the head of three is a shape move that jumps to the eye of any experienced player. Being in the center of the board, not only it's a boundary play between Black's center and White's right side, it also inmproves the health of Black's strings that surround that center, reduces that of the 3 white stones, affect the connectivity at the bottom and threaten to cut through White's whole structure at the side. It even unlocks some aji of the marked captured stones.

Such moves are easier to find than a double sente like A. They require no reading, only visual recognition. It is very hard to quantify such moves though. Center territory is always smaller than it looks, while central strength is more important than it may seem at first. How to offset them agsinst a 12 point gote at G, which also affects the life of 2 groups? As we will see, it's best for Black to take this move first and grab G next.

Related endgame areas

Although we found that in this case it was safer for White to respond locally, we analyzed that the circled points in the center were a possible response to the squared move and vice versa. This is not just mutual damage. The idea is that two different endgame moves affect the health of a group, so rather than responding locally, in order not to die, it can be better to move into the other area, finding life and reducing opponent territory at the same time.

The fact that this miai endgame exists, reduces the value of each move. Even if both are locally sente threatening, overall the position remains balanced and hence neither move is urgent. In this case, since the squared move was more severe and moving into the center was less secure, it became more urgent.

When White's right side has been reduced from the center and strengthened in the process, the urgency of increases. The value of a move is equal for both but the play alternates, so here Black has found a way to induce value for himself.