4-4 point 3-3 invasion double hane
Return to parent joseki article: 4-4 Point 3-3 Invasion, Hane
The 4-4 point 3-3 invasion, double hane initiates a popular contemporary joseki played after the 3-3 invasion. It is a territorial move that aims to take back the corner, but it cedes sente and influence in the form of a ponnuki to the invader. In the AI era, it is considered to yield an even result for both players on an empty board, but traditionally it was thought to be more of a situational move. In 2019, it was the fourth most common joseki played in professional games.
|Table of contents|
Moves are listed by frequency in professional games, which is sensitive to whole-board position. Bolded moves are commonly considered joseki.
- a, 4-4 Point 3-3 Invasion, Double Hane, Atari Up - (joseki) (common) (beginner) (AI favorite)
- b, 4-4 Point 3-3 Invasion, Double Hane, Counter Hane - (traditional) (situational) (intermediate)
- x, 4-4 Point 3-3 Invasion, Double Hane, Connect - (rare/mistake)
- y, 4-4 Point 3-3 Invasion, Double Hane, Atari Sideways - (rare/mistake)
|Atari Up (~81%)|
- The atari up with is the most common continuation and it is generally considered correct by AI (Diagram 1). On an empty board, this joseki is considered even for both players. White accepts that Black will capture the corner, but White will obtain a ponnuki and end the joseki with sente. Black also retains some bad aji, so White's continuation at a is potent. Furthermore, White can play at z to threaten saving the captured stones.
- Historically, it was possible for Black to atari at to build thickness while threatening a ko (Diagram 2). Black hopes that White will connect with at to avoid a ko. However, AI generally believes that White should counter-atari at regardless of ko threats. White does not need to fear a ko because (A) there no ko threats in the opening and (B) Black must invest several moves to start and close the ko, which is usually too slow if Black attempts to resolve the ko immediately. Jiang Weijie calls a "very bad move" in the AI Weiqi Joseki Dictionary (2021), and it is virtually unseen in contemporary professional play.
|Counter Hane (~19%)|
- The counter-hane with was historically considered a joseki that White can play to switch the direction of the ponnuki. However, it is no longer considered joseki in the AI era, as it is believed that White lost too much. Importantly, Black's position has little aji, so the result is considerably better for Black than the variation with the conventional ponnuki.
- In the old joseki, extending with is the traditional move (Diagram 1). This variation is better for endgame, but it permits some outside aji at a or b.
- AI often prefers to atari with (Diagram 2). This makes Black thicker and eliminates the outside aji, but White will have better endgame at a. This variation is recommended for general purpose use in the AI Weiqi Joseki Dictionary (2021) and by several contemporary professionals.
- It is possible for Black to atari with (Diagram 3) to pursue a fighting variation if they have friendly stones on the top side of the board. Without support (or if the support is misplaced), the atari is often inferior to the solid connection at m.
- It is extremely rare for White to play for the corner territory after the double hane, although it is possible for White to live small on the inside, often in gote or at the expense of ceding too much thickness. In most circumstances, these variations are better for Black. Many joseki dictionary mark these variations are mistakes.
- In order for White to live locally, they must create a L+2 group or a J+1 group.
- In some circumstances, professionals have played the hanging connection at a instead of the solid connection (Diagram 2 / 3). The hanging connection allows White to take sente, but Black can turn the corner into a ko.
yuzukitea: The consensus among professionals and many strong players is that that "switcheroo ponnuki" (coined by Uberdude) is rarely ever good. Zhan Ying (2p) states that "after extensive research, it seemed as though there weren't any situations where the switcheroo ponnuki was good". The lifein19x19 forum thread " Is the 3-3 double hane switcheroo ponnuki joseki ever good?" appears to have reached a similar conclusion. Aesalon? said that they went on multiple quests loading ~100 professional games in attempt to find situations where the switched ponnuki isn't a bad move, but they are extremely rare in professional play. This is especially the case in contemporary go, since 3-3 invasions are usually played relatively early in the opening.
In most cases, AI considers the counter-hane at a to switch sides to be a mistake. On an empty board, changing the ponnuki's direction is a five point loss in estimated score. It is extraordinarily rare for the alternate ponnuki to be worth the five points it loses.
Zhang Qiang vs. Zhao Chenyu (2019) is a rare example of a game where AI recommends the counter-hane at a to switch sides. In this game, we can see that the regular ponnuki arising from b would be extremely dissatisfactory, and there's hardly any way to imagine how it would be good by any stretch of the imagination.
Since b is eliminated as an option, White has no choice but to play the counter-hane at a. White is not precisely happy to switch sides (it is variation that inherently loses ~5 points), but in this case the conventional ponnuki is even worse. When presented with two bad options, you must pick your poison.
yuzukitea: Generally speaking, the outside ponnuki is better for White than the corner, but it is important to note that it is possible for White to live inside the corner if they so desire. There are in fact multiple methods for White to live small on the inside (often in gote), and several variations have been played by professionals throughout history.
The simplest move to live on the inside is to connect at b, but Black becomes extremely thick and it's generally a bad result for White. Other potential moves that live on the inside is hane at d followed by a solid connection or hanging connection (for a ko). White can also cut at c followed by the hane at d. (see: BQM342)
I struggled to find professional games where AI agreed with the professional move to keep the corner. In most cases that I examined, AI disagreed with the historical move.
Wu Guangya vs. Tang Weixing (2018) is a rare instance when AI agreed with White's decision to keep the corner. Here, the best option is for White to atari at a followed by connect at b. In this board position, Black cannot ladder the stone and they also cannot save the stone. Furthermore, if White continued with the regular ponnuki, they would be extremely overconcentrated due to White's nearby living shape on the top side.
While it is generally a poor idea for White to play for the corner territory, there could be extremely rare situations where a non-standard move is good.
-  Frequency statistics were obtained from Waltheri's Go Pattern Search using the full database restrained to a local search (accessed September 2022).
-  Online Go Server Forum: The Twenty Most Popular Current (2019) Joseki (2019)
-  AI Weiqi Joseki Dictionary - Jiang Weijie (2021)
-  (Bilbili) Zhan Ying: AI围棋定式3-最受欢迎的点三三应用秘籍 (2021) (CN)
-  (Youtube) Ji Lili: 星位 点三三（上集） (2021) (CN)
-  (Youtube) Michael Redmond: Learn Redmond's AI Style Go in 5 minutes - ALL About the Direct 3-3 (1) (2020)