Share your games with 6+ groups

We already have threads for sharing shapes, seki, … I was reviewing a couple of my recent games from last week, and found one where I won with 6 living groups, in defiance of this proverb. It was actually interesting to review with this in mind, because the AI seems to think that the game turned in my favor around when both myself and my opponent were ignoring my weakest group, which eventually was able to live.

link to the game

I’d love to see some more examples! Victories or defeats, 6 or more living groups, anyone have anything interesting?

(I looked to see if there was another thread but I didn’t find one, if this is a duplicate, my apologies)


I remembered a Sensei’s Library page with an archive of games like this.

So I went to Five Groups Might Live But The Sixth Will Die at Sensei's Library

On that page, I found the comment

For anyone else reading this, do you remember a page of examples of players winning with six or more groups? I can’t find it. Thanks; – bugcat

and the footer

Five Groups Might Live But The Sixth Will Die last edited by bugcat on August 26, 2020

Somehow I don’t think there’ll be any luck on that front coming soon.


See also


Takemiya Masaki vs O Rissei (40th Judan, game 4, 2002-04-18)

Black loses by 1.5 points with seven groups (5.5 komi)

seven groups win game.sgf (1.5 KB)

From Not So Magnificent Seven at Sensei's Library


I couldn’t find any six-group games, but I did uncover a five-group 20-point win I got against @alemitrani back in 2018.

The first group founded was :bug: with (2) C4. I must have been in an interesting period of my opening development because I played (4) as a corner enclosure.

The next, unnumbered group was O17 with (6), which as you can see eventually died. However, its extension at (9) K17 survived and become the :cat2: group.

The third group was :tiger2: with (12) R6, approaching the lower right.

Black’s (19) M17 split :cat2: from the O17 stones.

With (20) O17, invading the corner, I started another group; this group also died, but by (32) its southern remnant had conglomerated into :tiger2:.

(36) F17 was initially an extension from :cat2:, but it turned out to be the foundation of :cricket:. With (41-5) Black split those two groups.

During the running fight on the top side, :cricket: was actually able to connect on (76) to :bug:, forming the supergroup :bug::cricket: which lasted for about another 75 moves. At this point, then, there were three groups: :bug::cricket:, :cat2:, and :tiger2:.

On (98), I tried to break into Black’s central territory with L4, the root of the final group, :beetle:. Black split with (101); however, after the fighting was resolved I still had the opportunity to connect to :bug::cricket:.

For a time it was possible for :cat2: to connect to :tiger2:, but Black prevented this with his split on (137).

On (156) I apparently tried to play some very fancy endgame, but all that came of it was that :bug::cricket: was resplit on (159).

On (202) I committed to the separate existence of :bug: and :beetle:, defining the final five groups.

I might easily have finished with only two groups, :bug::cricket::beetle: in the west and :cat2::tiger2: in the east, but in the game I ended up, as you see, with five.

By the way, it’s Firefox that has the cute pixelly emoji. I wish Discourse retained them.

9.5-point five-group win from September.

In that game it’s obvious that I took four corners, lived on the right side and invaded the centre twice, fleeing to the north and south.

To take four corners and win, you typically do have to finish with at least five groups.

4.5-point loss from 2018, five groups.

As you can tell, I haven’t found a sixer yet.


Great idea of presentation! How did you do that, by hand or something automatic? If by hand you’ve given me an idea for a new feature everyone secretly wants :star_struck:

It’s also interesting to think about the age of the groups… I guess that’s an advantage to the style of kifu where every stone is numbered

1 Like

It’s automatic.

tl;dr you select the letter (A) tool, hold Shift and click on the board to open a custom label box.

Specify the label and away you go~


Five-group 10-point loss. Really four and a half groups since I can connect at P8.

Five-group 20-point win.

Five-group 10-point win.

Five-group win by 3.5 points.

Five-group by 5.5 points.

I haven’t been able to find a single game I’ve played in which six groups of either colour lived. It seems to be a much rarer occurrence than five groups living.


Won many games with 4 corners+ center.
In that kind of game it’s key to: take care of the center, may die easely so don’t hesitate to add moves there. Restrain to go for edges. Enough is enough.

1 Like

Just finished this one. Handicap = 3.
Just kept on thinking about:
six groups, one dies.

After 100 moves

After 200 moves

Final situation


I found the emoji diagrams so amusing but I’m allergic to clicking the mouse that repetitively… so I modified @yebellz 's Besogo to make it easier. link Just pick your emoji and then click a stone and it should mark all connected friendly stones (not doing any kind of smart analysis so no diagonals)

This is a 9 stone handicap game against my friend (chess player but go noob) that I remembered… I kind of feel like this only half counts (yeah… I missed that those cuts by the space invader group, so black won)


I’m allergic to clicking the mouse that repetitively

Click-and-drag is supported.

1 Like

It’s a long story but dragging is actually harder for me :sweat:

1 Like

It could be a cool feature to be able to hold another button down, and click and have it mark the whole group with a number/letter/triangle/circle/custom label

I finally found one!

This is my 14.5-point loss as White against fromtasmania, from October 2017.


(2) – :wine_glass: is founded at R16.
(4) – :beer: is founded at Q3.
(16) – :sake: is founded at C11.

(23–31) – Black splits :beer:, the western part becoming :beers:.
(32) – :cocktail: is founded at C17.
(68~78) – :wine_glass: expands west.

(79–91) – Black separates Q10 from :wine_glass:; it grows and is absorbed into :beer:.

(167–176) – I make a bad endgame mistake, allowing Black to once more split :beer:, which has to relinquish the Q10 stones into the independent group :clinking_glasses:.

This is an unusual six-group even game because I only took a corner once.


An interesting thought is that if only five groups can usually live, that means the typical group requires between 30 (six groups) and 36 (five groups) moku as minimum living space.

The average, or “minimal bubble”, is 33 moku.

So, is it 33 moku on all larger boards? Or does the number grow larger or smaller?

What about on the smaller boards?

(As a side note, I tend to use moku to refer explicitly to intersections rather than “score points”.)

1 Like

Does this assume a relatively close game by score? Just trying to wrap my head around the analysis… Does the amount change depending on the number of opposing groups?

I think it’s interesting that unless I’m mistaken all the games posted so far the opponent has a single connected group… (in the pro game there are two groups, btw I think white won that right thus not so magnificent seven?)

The disparity is interesting, but I don’t think I’m strong enough to make sense of it

There are 361 (call it 360) moku on the board and each player can, by the proverb, make between five and six groups (or more mathematically, 5 <= x < 6).

That gives us 360 / 10 = 36, and 360 / 12 = 30, the average of which is 33.

I think white won that

The game legend says that Black won.

I think 9x9 games usually end with two or three groups total. And 81 / 2.5 = 32.4, pretty close!


Weird I was confused because the discussion on SL someone said black is winning on the board without komi but white won by a point or something… I haven’t bothered counting

1 Like