4-4 Point Traditional Slide Joseki, 2-8 Invasion

  Difficulty: Advanced   Keywords: MiddleGame

This page is in need of attention.
Reason: Original research using Waltheri / AI and personal inference. Would be nice to have corrections from stronger players.

Return to parent joseki article: 4-4 Point Traditional Slide Joseki

2-8 Invasion  

The 4-4 Point Traditional Slide Joseki 2-8 Invasion refers to a middlegame pattern that often occurs around the traditional slide joseki when Black has an additional approach stone (marked circle). This invasion is aggressive and seeks to take away White's base or split White's group. It may lead to complicated fighting.

Responses to this invasion have evolved over time. Strong AI programs have different opinions about this invasion, and professionals has historically played sequences that are not viewed favorably by AI.[2] This page attempts to document the progression from historic middlegame joseki to current AI-inspired thinking.

Patterns around this position are highly sensitive to whether the Black approach stone (circle) has support. If Black has a strong position on the top side, the ensuing fight could favor Black. Without support, strong AI programs tend to favor White in the fight.

When Black is weak, Black can also treat B1 as kikashi to build thickness on the outside.

Table of contents


Blocking at a is chosen by professionals a majority of the time (~78%) [1], and it is White's strongest response.

Blocking at b is passive and allows Black to connect up to the outside, but it ensures that White's stones will remain connected. Sometimes Strong AI programs can prefer blocking at b if Black is very strong in the local area, which would make a potential fight unfavorable for White.

Diagonal Attachment Variations

Clamp and Fighting Variation

Kim Eunseon (3p) vs. Rui Naiwei (9p) (2007) - Colors Inverted  

In the mid-2000s, one of the main variations considered by professionals was the diagonal attachment B1 followed by the clamp B3. This clamp was played when Black has a strong group at the top side (marked square) and seeks to separate White's stones.

A match by Kim Eunseon (3p) vs. Rui Naiwei (9p) (2007) illustrates this concept. After the clamp at B3, the following sequence of moves are the strongest resistance by White and in fact recommended by AI.

At first glance, White appears to be in a bad position because White is split into two weak groups while Black only has one weak group to manage. In the ensuing fight, Kim Eunseon (white) struggled to recover from this position and was unable to save both sides. Professionals seemed to dislike this variation and it was common for White to play moves to avoid this complicated fight.

The subsequent variations listed in this section are countermeasures against the clamp and methods to simplify the position at a local loss for White.

Clamp and Submissive Variation

Mukai Kaori (1p) vs. Yamamoto Kentaro (4p) (2006) - Colors Inverted  

It is possible for White to decline a fight and remain connected with the hane at W4. This variation is depicted in the book, After Joseki, although it is almost non-existent in professional games. Mukai Kaori (1p) vs. Yamamoto Kentaro (4p) is the only result in Waltheri that matches this pattern.

Clearly, this result is too good for Black, and Black even takes sente.

White cannot accept this variation.

Extend Variation (Mid-2000's Joseki)

Lee Changho (9p) vs. Lee Sanghun (7p) (2002)  

The most popular variation played by high-level professionals in the mid-2000s was the extend W2 in response to the diagonal attachment B1. The apparent purpose of W2 is to prevent the clamp (discussed above), and it aims to simplify the position in sente while keeping White's stones connected.

Black takes solid territory with the descent B3 and White remains unsettled. The black+circle stone still has aji, and the position seems better for Black. Nonetheless, this appears to the most common variation played by top professionals prior to the AI revolution, and it is seen in the games of Lee Changho, Gu Li, Takao Shinji, Kong Jie, and others.

Strong AI programs do not evaluate this extend favorably, since White gave up too much too easily. In fact, AI seems to believe that fighting (discussed above) is better for White in most situations. However, strong amateurs like Baduk Doctor? (7d) favor this extend variation (see linked [ext] youtube video).

Kikashi Variation

Kobayashi Satoru (7p) vs. Cho Chikun (9p) (1984)  

Black can choose to sacrifice the invasion stones to build thickness on the outside in a form of kikashi. This follows the go proverb: "Add a second stone and sacrifice both".

The following sequence of moves depicted in Kobayashi Satoru vs. Cho Chikun are nearly all forcing and recommended by AI. Black's descent at a is forcing and b nearly captures W8. This variation is decent for Black even if Black does not have the ladder.

Strong AI programs tend to prefer this variation over the diagonal attachment and clamp.


Strong AI criticize the extend variation

Strong AI programs [2] dislike the extend variation (discussed above) and prefer that White block and fight (discussed above). Although the fight may appear difficult for White, the AI favors White in the fight, since Black has poor shape with a shortage of liberties. A hane at the marked square takes advantage of Black's liberty shortage and White has many forcing moves in the center.

From an amateur's perspective, it is difficult to assess whether human players would be able to manage this fight as well as strong AI. There are no professional games in Waltheri illustrating a fight at that went in White's favor, and these variations have not been played by professionals since the AI revolution.

See also


  • [1] Frequency statistics were obtained from [ext] Waltheri's Go Pattern Search using the full database restrained to a local search (accessed August 2021).
  • [2] AI score estimates were obtained from katrain (v1.9.3) and sabaki running 30 block katago (1.8.0). AI score estimates can vary widely depending on the board position and software version, and should be taken with a grain of salt.

4-4 Point Traditional Slide Joseki, 2-8 Invasion last edited by yuzukitea on August 17, 2021 - 01:04
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Login / Prefs
Sensei's Library