Go Wizardry

All About the Many Aspects of Go
We have millions of friends around the world... and they all play go!



早乙女や朝澄む小田の水かゝみ        蕪村
Saotome ya sumu oda no mizu kagami      Buson
Rice planting girl! The mirror of the water of the clear rice paddy
Artist: 深林人 Shinrinjin

Kido, July 1974

Yosa Buson (与謝蕪村 1719-1783) appeared on the cultural scene (as both a poet and artist) after the great Matsuo Basho (松尾芭蕉 1644-1694) had passed away and the haiku had suffered a decline. Buson’s haiku displayed a grace and elegance that breathed new life into the art. There is also a vivid artistically visual sense conveyed by the work. In the above haiku, one can almost hear the girl singing traditional rice-planting songs as she works. Buson’s published masterpiece is 「新花摘み」 ("Shin Hana-Tsumi" = "New Flower Plucking").

It has been suggested here that the way to challenge AlphaGo in the most effective way possible is to play mane-go, or "imitation go," against the AI program. In fact, in terms of pure theoretical science, AlphaGo could play mane-go against itself to determine the precise value of the komi needed to be applied to make the game as equal as possible. It is interesting that although the komi has been steadily raised over the years, Black’s winning rate has stayed almost exactly the same! That is, Black wins more than 50% of the time, with the rate varying between 51% and 53%.

I remember years ago I was at the Cotsen Open and complained to a friend that I was always given more games to play with White than Black. This is frustrating, because I was playing in the top band, against the strongest opposition, and having to continually to play defensively from the start was wearing me down.

"But look at this!" my friend replied, grabbing eight Black stones in his hands. "Look at the size of this komi! What an advantage!"

In the next round, my friend played a Chinese semi-professional player and lost the game by resignation. In the latest Cotsen Open, he played another Chinese semi-professional, and lost by 1 1/2 points. So much for his theory of the advantage for White with a 7 1/2 point komi…

I doubt that it is possible for even top professional players make a definitive judgment as to the exact value of the komi. So in the interest of advancing research, I offer the following analysis of mane-go in order to stimulate discussion of this.

From Kido, September 1975



"Anti-Mane-Go Strategy?"
Those interested in replaying the game move by move can click here to do so.

Honinbo Tournment, Preliminary Round Kajiwara Takeo (Black) / Fujisawa Hosai

According to Kajiwara 9 dan’s words, "A move at the tengen center point has too weak force against the corners, so a move there does not work with 100% effectiveness." But in this game, he violated his own precept and made his first move at tengen. This was probably a ploy to prevent Fujisawa Hosai 9 dan from using his favorite mane-go technique. In the past, Kajiwara has played at tengen against Hosai 9 dan, so this was the third time. Regardless of that, he takes the wins, which is only to be expected [of someone of Kajiwara’s strength]. In this game, too, through Black 41 it may be thought that the play is proceeding at Kajiwara’s pace.

From Kido, March 1974



"Mane-Go Destruction, Two Years Later"
Those interested in replaying the game move by move can click here to do so.

Meijin Tournment Yamabe Toshiro (Black) / Fujisawa Hosai

If the reader is thinking that this board position somehow seems really familiar, it is because in the semi-final of the Nihon Ki-in Championship Tournament two years ago in 1972 [the following game] the same competitors played the same opening. In that game, with White 2 and Black 3 in place, Black played this move at 1 at "a," and White responded at 4. Yamabe won both that game and this one, displaying a model to use to defeat mane-go. Please read the analysis in the Yomiuri Newspaper by Yurakushi for the details.

From Kido, January 1973

Successful Destruction of Mane-Go

20th Annual Nihon Ki-in Championship, Semi-Final

White: Fujisawa Hosai 9 dan

Black: Yamabe Toshiro 9 dan

Played in November 1972 at the Nihon Ki-in, Ichigaya, Tokyo.

Komi = 4 1/2 points

149 moves. Black wins by resignation.

Analysis by Yamabe Toshiro

The Mane-Go Conception


Figure 1 (1-27)
Those interested in replaying the game move by move can click here to do so.

Black and White both play the three-star points in a row opening.

In response to the attachment of Black 9, White plays 10, launching the insipid mane-go strategy.

As is well known, this is Hosai 9 dan’s treasured ploy, and at this point I had lost more than ten straight games against him.

If a win or two could be entered into the series, this kind of record could be said to belong to a "duck" [鴨 = kamo] or a "customer" [お客さん = okyaku-san], but when this kind of drubbing continues for so long, it becomes a question of whether the ranking conditions are correct. In this regard, I had come to doubt myself.

What is more, when I was playing Black, almost all the time I would be faced with mane-go, and despite this I could not come up with any effective counter-strategy. Along with that, the embarrassment of insufficient study and research was added into the mix.

Well then, instead of the attachment of Black 9…


Diagram 1

…I also considered making the pincer of Black 1.

If White obligingly continues mane-go with 2 and the following moves, occupying tengen with Black 15 ends the mane-go and this move represents a shining light that works effectively for both attack and defense.

However, this depends on the mane-go continuing, the strategy based on that premise, and the fact that this will not necessarily happen is the source of uneasiness for all who face mane-go.

Therefore, I chose to make the attachment of Black 9 so as to adopt a simple way of playing.

The attachment of White 10 and the following, in the local context, is an odd progression of moves, but this kind of symmetrical play is typical of mane-go.

Instead of the atari of Black 27…


Diagram 2

…if Black extends at 1, White extends at 2, and the following sequence through the turning move of Black 9 may be considered.

When Black plays 11, if White ends the mane-go with 12, supposing that White backs off this way, after White 20 it comes down to a question of territory. Under this assumption, the Black territory within the outline of x’s is equal to 21 points, while the White territory is equal to 16 points.

That difference of 5 points is insufficient to overcome the komi when it is 5 1/2 points.

Of course, this variation is flawed in its structure, but the result of various trials indicate that "ending mane-go" boils down to the sole question of territory, and in that case it may be thought that one way or another Black cannot win.

Black plays the atari of 27 so that rather than becoming a question of territory, the mane-go turns into a fighting game. That way, when the mane-go ends, a decisive advantage has already been established.

That said, showing the concrete execution of how to achieve that decisive advantage must be deferred to the next figure. After racking my brain over the matters described, I came to my humble conclusion.

Foiling Mane-Go


Figure 2 (27-48)

In reply to Black 27, White plays atari at 28. Here, trading capture for capture to end the mane-go is a difficult way to proceed, so it is natural for Black to connect at 29.

White extends at 36, but this is another turning point.


Diagram 3

The capture of White 1 ends the mane-go, but of course as a result Black makes the ponnuki one stone capture at 2. [Conf. the go proverb: "A ponnuki is worth 30 points."]

This mutual capture from a commonsense perspective is better for Black. However, something that is a little troubling is that Black’s territory on the lower side is too wide. When facing the attack on the corner of White 3, it is difficult to find a way to handle the situation.

Since Hosai san did not choose this way of proceeding, naturally he must have disliked this continuation. But in that case, the mane-go keeps on going.

The reason that Black inserted the cut of 39 into the sequence is that following White 46, for Black 47…


Diagram 4

…it was expected that Black would proceed by playing at 1.

Supposing that White obligingly continues the mane-go with 2, in the following sequence the same shapes are produced until Black cuts at 15 to bring the mane-go to an end.

Should White cut at 16 and 18, after Black 19, White is destroyed. In other words, in the course of the fighting, Black can be played at any time as a forcing move. (If White uses 16 to simply cut at 18, the hane of Black 19 is tesuji, and White does not have the resources to fight.

On the other hand, in the game, Black played differently, extending out with 47. That is because…


Diagram 5

…if Black plays at 1, I worried that White would suddenly change direction and come at Black with 2 and 4. In this diagram, incurring the diagonal attachment of White is bothersome, so the fact is that I ended up getting scared.

Supposing that pressing with Black 1 in Diagram 4 works out well, with the stones played with no laxity so as to be a "mane-go counter-strategy," I could have felt proudly self-satisfied, but in the game the move of Black 47 is lax, so I feel little pride in it.

Without a doubt, there is certainly a reason behind Black’s move.


Diagram 6

Should White imitate Black’s move in the same way with 1, I intended to jump with Black 2.

The jump of White 3, rather than being an imitation, is played because White has no other move to make than this one. Black then pushes through with 4 and cutting at 6.

With this, viewing the state of the position, it may be said that it is clear that White has no ability to fight.

Consequently, White comes to press at 48, ending the mane-go. This is, first of all, an inevitable development, and without doubt the mane-go counter-strategy has been successful, but the success or failure is still to be decided.

Running a Risk


Figure 3 (49-69)

Moving out with Black 49 and 51 is unavoidable.

In response to White 52, Black 53 is a natural defensive move. Suffering a fencing-in move [at the point below 53] and being squeezed would have hit home hard. [Explanation: After White plays at 52, the ladder with White 53 does not work, so Black might be tempted to play elsewhere, but White then makes a fencing-in move at the point below 53. Black can move out at 53, but White plays atari at the point to the right of 53, then squeezes with an atari at the point to the right of 52. When Black connects, the Black stones are a lump with little operating value, while White has gotten thick and strong in the center.]

For White 54…


Diagram 7

…after the game, Fujisawa said that he considered attaching with White 1, but in that case play develops with Black 2, 4 and 6, and it is unclear which side benefits.

Instead of playing at 55, Black should probably have desperately went to make life life on the upper side, but while playing there White would be given a chance to play at , and it seemed to me that the scope of battle would be enlarged.

Black 55 is met by White jumping to 56, whereupon I was running the risk of things becoming difficult and painful in the upper area, but regardless of that, I wanted to divide White’s stones to secure a footing from which to attack.

Black plays 57 and the following moves in order to survive [shinogi], so that at the point that the mane-go ended, suddenly a violent phase of the game was reached. Since to a certain extent the shapes on the board have been set, the scope for variations has become small, so there is no alternative.

With White 58 and the following moves, the upper side steadily becomes solidified, but this is an unavoidable progression of events. In response to White 68, Black attaches at 69 to switch to counterattacking, giving Black 55 a major role in the action, which fit in with my intentions exactly.

Large Prey


Figure 4 (70-94)

From the hane out of White 70 to White 76 is an unbranched path.

The wedging-in move of Black 77 is the only move. Extending at the point of 78 would incur White’s connecting precisely right at 77, which would be no good.

Capturing Black’s two stones with White 78 and 80 is unavoidable since blocking with White 79 would be no good.

Capturing White’s two stones with Black 81 in sente is extremely thick and strong.

In reply to Black 83, White desperately pushes through at 84 and cuts at 86.

The atari of White 88 and the move-in-a-row of White 90 are the best and strongest moves.

When Black plays at 91, the peep of White 92 is makes Black’s group into a large prey being hunted. During the skirmishing in the center both sides have captured two stones, but Black got the better of the trade. On White’s side, the eye shape cannot be said to be perfected. Therefore, it is natural for White to initiate a desperate, all-out attack, one that stakes the outcome of the game on this fight.

For Black 95…


Diagram 8

…hastily making an eye with Black 1 is no good.

This incurs the moves of White 2 and 4, and if Black plays 5, the hane of White 6 and connection of 8 make it impossible for Black to make two eyes. In particular, incurring White 4, capturing a stone, means that the potential [aji] of a cut at has been lost, which is terrible.

Here, there are other possible variations, which will be touched upon a little in the following figure.

No Eyes


Figure 5 (91-122)

Instead of descending with White 94…


Diagram 9

…should White hane at 1, the cut of Black 2 is tesuji. Connecting with White 3 is standard, and then Black 4 is a skillful move. White 5 is met by Black playing 6 as a sacrifice stone, then Black 8 and 10.

White has no choice but to jump at 11, so Black pushes in at 12 and cuts at 14. This produces effective moves.

Even if Black plays simple moves like 16 and 18, following the sequence through White 23, Black is a forcing move that lets Black live with .

Once the group is alive, this game is over.

Therefore, instead of White 7…


Diagram 10

…White might go after Black with 1 here. Then, Black plays 2 enticing White into filling a liberty with 3.

When White takes Black’s other eye with 5, the throw-in of Black 6 allows Black to move out with 8 and 10. The block at and the potential [aji] of a cut at ensure that Black can make a second eye in the corner.

For that reason, it is unavoidable that White play at 96 to take away the eye here. For Black’s part, the jump at 97 aims at making an eye in sente.

White also must jump to the point of 98. There are potential problems [bad aji] here, but there is nothing to be done about that.

Black 101 and then 103 is the correct order of moves. No matter what Black does, this big group of stones only has one eye, but these maneuvers are undertaken to extend the group’s liberties.

The cut of Black 105 is made at a difficult move order point, but at this stage, White has no alternative but to answer at 106. When Black goes back to play at 109, it is natural for White to play at 110 to take away the eye. The Black group ends up having only one eye.

Black 113 spurs White on to playing 114 and 116, a method adopted to put the stone of Black 115 in place. In short, this in fact threatens White’s eye shape in the center. And when play reaches this point, there is a race to capture with the center.

Black 117 and 119 are also natural moves. Black does this to extend the number of liberties of the group, then connects with Black 121.

White 122 is unavoidable. White wants to go back to play in the center, but…


Diagram 11

…after incurring the peep of Black 3, things will not go smoothly. Should White reply at 4, Black then play at 5, and if White . Black . The atari of Black can be made at any time as a forcing move that must be answered.

It has come to the point of eat or be eaten, but I could see the way through to the end.

Attacking and Winning


Figure 6 (123-149) White 26 connects

Being able to make Black 27 as a single forcing move is big. By making the incursion here, even if the Black group is captured, White will be wrapped up precisely and squeezed, and the position on the board is such that Black could aim to with the game by way of territory.

Black 29 makes it a race to capture, but instead of this move…


Diagram 12

…the knight’s move of Black 1 is simple and easy. With White 2 and the sequence through Black 15, Black gives way move after move, ending with the win. If White , Black throws in at , and White cannot make two eyes. [Not true: White makes one eye, then, after the Black throw-in at , White answers at the point to the left of Black 1, and after Black captures White’s three stones, White captures Black’s stone below 1 to make the second eye. The writer working with Yamabe must have made a mistake here.]

Playing White 30 at 33, followed by Black , and White 35 would have made things more difficult. But since White plays at 30, with the moves through Black 45 the race to capture is won by Black.

The cut of Black’s marked stone works effectively, giving Black’s group more than ten liberties.

149 moves. Black wins by resignation.

Those who wish to comment on the opinions expressed here may send their thoughts to info@GoWizardry.com. The most interesting responses will be addressed in future postings.

Robert J. Terry

Tagged as: , , , , , , ,

Leave a Reply


book cover

Go on the Go Collection: Volume I

Three booklets have been assembled into the collection here.

Buy this Book at Amazon

Go For Everyone

Go For Everyone

A New Method for Learning to Play the Game of Go

Buy this book

Book Cover

Journey to the West

This is a semi-autobiographical novel that depicts a unique American success story; a rags to riches tale of a man escaping his humble origins to make millions of dollars, but then he throws it all away due to the ancient character flaw of hubris.

Buy this Book at Amazon