Endgame Tesuji 1

I am a big proponent of learning to play the endgame well, and I actually consider my endgame to be the strongest part of my game (not implying it is strong, though). I know absolutely nothing about the opening, and my endgame has on occasion turned a potential one-sided beatdown into a close game.

Much like the midgame, tesuji feature prominently in a good endgame. Clearly, there is more to the endgame than tesuji; specifically, counting and evaluating the order of endgame moves. However, it is tesuji that this post will focus on.

Figure 1 presents the position we will consider. It is adapted from a game between nover [5k] and takeko [6k] on KGS, played on August 24, 2013. Black is to play first, and has to find the best local endgame move.

Figure 1: Original position

Figure 2 below shows Black's most obvious move (when one is not thinking "tesuji!"). Black 1 is not a bad move, but it is not the best. In fact, Black 1 in Figure 2 is currently a zero-point move. We note that, in this position, White ends up with 10 points locally (9 points of territory + 1 dead stone).

Figure 2: The 'obvious' endgame move for Black

Figure 3 shows an optimal move by Black. If White tries to resist at 2, Black plays at 3, putting two groups in atari. White cannot capture Black 1, lest Black plays at 'A' and captures the 8-stone White group. White's best resistance at this point is to capture the triangle-marked stone. Black then plays at 'A' for a big ko in the late endgame stage.

Figure 3: The endgame tesuji and White's futile resistance

The proper response to Black 1 is to atari from the other side, as in Figure 4. Black 3 is then another tesuji to reduce White. In this position, we see that White ends up with 8 points locally (6 points of territory + 1 dead stone + 1 captured stone). By comparing Figures 2 and 4, we can deduce that Black 1 below is a 2-point gote endgame move.

Figure 4: The endgame tesuji and White's proper response

The example above may seem pointless. Black does 2 points better - whee! Indeed, in the endgame, each proper move is perhaps only slightly better than the more obvious, yet suboptimal move. Play a proper move consistently throughout, however, and they tend to add up. If we imagine a typical late endgame stage is about 50 moves long, then it is not unreasonable to think that proper endgame can be 10 points better than the 'obvious' endgame. In a close game, 10 points is certainly quite significant.

Black actually has more than one proper move in the position shown in Figure 1, which is why I called Black 1 in Figure 3 "an optimal move" rather than "the optimal move". All of the optimal moves lead to very similar positions, though, and exploit the same weakness in White's formation.