InvincibleTheGamesOfShusaku/Game 35 analyzed with KataGo

In the 6th Castle Game documented in Invincible, Shusaku takes Black against Yasui Sanchi. Commentator for the book is Sanno Hirotaka. The game starts off with a fight in the upper right.

Moves 1 to 10  

Obviously the fight stems from a taisha joseki.

Moves 11 to 20  

Sanno says B13 is no longer the standard move but the joseki of the time of commenting is also no longer valid by modern insights. Even the whole taisha has actually disappeared.

Also W20 is "the wrong side of the corner". Sanno recommends 4-4 or 3-3 instead. The reason is that the top is influenced by Black and White should emphasize the corner.

Book proposal for 13  

The bad looking empty triangle makes miai of W14 and B15.

AI proposal for 13  

The AI proposal is somewhat similar but puts more pressure on the opponent stones.

Book proposal for 20, locally  

Apart from criticizing the particular choice in the upper left, the book says the area is not urgent in the first place and it might be better to continue locally. AI doesn't like this so much, as it allows Black to play in the open corner(s) first.

AI proposal for 20, locally  

AI suggests this W20. Black needs to prevent an attack so W22 can occupy the empty corner.

Moves 21 to 30  

Sanno critizes W22 for being the wrong direction. KataGo locally prefers the outside attachment and globally continuing in the upper right or taking an open corner, but all of it by a small margin. The main reason why it seems to dislike W22 is because Black can now easily tenuki and White can't take the corner efficiently. From various local suggestions by KataGo it seems to value the top side more than anything. While Sanno thinks Black's influence is to stay away from, KataGo treats it more like a potential target.

Hence it approves of B29 but likes A a little better. KataGo would solve White's issues differently:


W4 is using the aji on the right side to move faster into the center.

Moves 31 to 40  

The fight continues. Sanno likes the flow for Black. KataGo dislikes W38: the focus is on the running fight into the center. A is the vital point. Hence W40 is an overplay according to KataGo.

Moves 41 to 50  

However, Black's bamboo is slow. It allows White to occupy the vital point with W46. Simply moving out diagonally would have been more efficient.

The book discusses at length the top right life and death. Moves in the vicinity of B49 are sente to that group because a ko can be launched. W50 ix kikashi before living. Sanno calls it exquisite timing. KataGo holds an alternative idea.

Sacrifice - B9 at W4  

KataGo indicates that capturing the upper right requires a lot of work, since White has a big eye. So it is possible for White to sacrifice the corner, playing the thick escape of W2, then allow Black to win the ko by taking a big point with W8. When B9 connects the ko, White has forcing moves at A and B. White could even exploit Black's bad shape with C and D.

This position is a good example of AI's extreme flexibility. It will continue locally as long as the stones can be used efficiently but as soon as the situation reverses, it will play elsewhere and sacrifice.

Moves 51 to 60  

If W50 had an interesting strategic alternative, B51 is plain wrong says KataGo. Black should simply connect with good shape at A. That would of course still leave the aji for the corner to escape but KataGo seems less concerned with that corner and more with the balance of power in the center.

Sanno calls W54 submissive but unavoidable. KG agrees White is falling behind but suggests fighting back at A.

Also KataGo disapproves of Black's tenuki. It's a common feature of strong humans to play away after having forced the opponent and obtain a favorable exchange like B53-W54. But KG indicates the whole board fighting is not over yet. Locally, Black can simplify at B, globally keep on the pressure with C or even cut at D. Once more we see how AI is not straightforward in its preference for open corners. As long as the local situation affects the efficiency of the stones involved, it prefers to continue. Black's tenuki at B55 amounts to a 4 point mistake!

W56 shows Yasui Sanchi's fighting spirit but both Sanno and KataGo think it's too much. KataGo prefers C again for White. It seems to be the vital point globally.

Moves 61 to 70  

B61 leads into a classic sacrifice tactic, allowing the extension at B69. Sanno applauds it, KataGo thinks White gets too easy life at the top. Also, B69 is pushing White over Black's weaker group, while building influence to the less interesting left side. Black's cut at A can now be met with White B, so there's not much of a threat going on there.

W70 is the point of the game where it gets close. This is a pattern in these evaluations of Shusaku's castle games with AI. It seems like Shusaku allows White to slowly get into the no komi game, to then widen the gap again from the late middle game onwards. This might point to Shusaku being relatively stronger in the late middle game but it might also indicate that he was able to build a position in the early stage from where it was easier to manufacture a win.

Interestingly, Sanno thinks the game is over after B69. Here the professional assessment of Black's overall superiority is not confirmed by KataGo's expected result.

Moves 71 to 80  

Sanno calls B73 a typical solid move when you are comfortably ahead. KataGo is not so sure about that: yes, B73 is lukewarm but no Black is not ahead more than 2 points.

Sanno also criticizes W76: it should be farther to the left, when playing at the bottom. KataGo thinks it should continue on the right side:


KataGo is more occupied with the relative strength of the groups than with developing the bottom.

Moves 81 to 90  

After both take big points before urgent ones, Black resolves the situation with B81. KataGo thinks White can fight back at B83 with W82 and should also unleash the aji higher up at A. Even after the sequence here White could do so.

W88 is not enough to win according to Sanno. He thinks White should invade at A, apopular probe.

alternative for W88  

A standard technique to set up a ko.

Moves 91 to 100  

Sanno thinks the game is hopeless for White by W94 but KataGo thinks it's still close. White could apply leverage on the lower left corner and threaten an attack on Black's top group. W96 at A could be such a starting point.

White chooses a more direct approach with W96 which Sanno assesses as a worthless peep, since B97 can push through. W98 backs off but KG thinks simply blocking was better.


If W3 blocks here, Black can indeed take sente but then White can make territory naturally while attacking the center.

Moves 101 to 110  

When B1 turns here, W2 loses tempo and White has to return at W4, so that Black comes out in sente and can reinforce at B5.

With that, it's indeed quite impossible for White to get back into the game. KataGo's evaluation at this point is B+5 There is still some aji in that top group but Black can live at A or come out at B, so there's not really a big problem for Black.

InvincibleTheGamesOfShusaku/Game 35 analyzed with KataGo last edited by Dieter on December 4, 2023 - 18:01
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