Capture go problems ⚫

I’m no expert in Capture go but can we infere that for experienced players, it’s like go, but with lower level of tactics (life is much easier) and finally the same goal (the one who control more wins as passing is not allowed)

C4 D4 C5 A6?

Ok nothing works. Other option is to start with this move A6, threatening A8

That is a good start! What happens after bA6 wA8?

I just get seki I guess then it’s incorrect

Depends on the rest of the board, who dies first? :slightly_smiling_face:

(one of the things I like about these whole-board capture go problems is that there is never any ambiguity about the goal. In regular go problems, beginners are sometimes confused about whether ko or seki are good enough solutions to a problem. For a capture go problem, as I’ve defined them here, the goal is to win the game. If you could play the position against an opponent and win every time, you’ve solved the problem. Otherwise keep looking. No ambiguity about what counts as “good enough” :smiley:)

Ah yes I see your point

Some extra explanation of problem 21:

Spoilers

Black needs to make seki between B4 and B2 to survive. To achieve this, preserving liberties is very important. Something straightforward like this obviously doesn’t work:

The key to the problem is that black can gain liberties by starting with the descent at A6:


This is the correct solution. (5 at C5 would also work, but black must not play 5 at A3 or B3. The mutual liberties are needed to make seki.)

Importantly, playing hane connect instead of descending is a mistake since it loses a liberty:


No way to survive for black after this.

Also, black can not start with the cut, since then white can resist with A7:


The upper left corner is locally dead, but it has four liberties since black needs to start with B9 to kill. So in this variation, black dies first since the black liberties were reduced by the 1-2 exchange. Thus the order in the solution diagram above is necessary: descend first, cut afterwards.

Please let me know if you have further questions or if you find any mistakes in my analysis!

3 Likes

Let’s switch things up with an easy one!

Problem 22


Demo board

By the way, feel free to share your own capture go problems here! To get started, play around with some random positions until you find one with a unique solution. It’s easy to make easy problems, like the one above. Harder problems are harder to come up with (duh). I would be delighted if someone could make one that I can’t solve :smiley:

1 Like

This is a great thread.

1 Like

Thanks @bugcat :blush:

I still hope some other people may want to try their hand at creating problems, so here’s another tip: just look through an existing collection of go problems and look for ones that also happen to work as capture go problems.

Below are two problems that were stolen inspired from Tsumego Pro:

Problem 23

image

Problem 24

image

(click on image to go to demo board)

The fun part is that only sometimes is the capture-go solution the same as the regular solution :wink:

3 Likes

Inspired by the recent discussion in this thread, here’s the same kind of question for capture go :smiley:

Is it possible for black to live inside this 6x6 area?

image
(click on image to go to demo board)

I think I know the answer, but I could very well have overlooked something, so I’m looking forward to other people’s input!

5 Likes
Partial spoiler (a failed attempt by black)

It looks to me like an invasion at the 3-3 point can be killed by invading at the 2-2 point :rofl:

Even though black gets some forcing moves…

…the extension at 11 can still be killed (just barely):

So assuming 3-3 dies, that leaves 2-3 and 2-2 as the remaining likely candidates to live. Hopefully someone else wants to take a crack at them, otherwise I’ll share another variation in a few days :slight_smile:

3 Likes
Some variations starting at the 2-3 point

Here’s a promising start for black:

This combination kills, but once again it’s just barely holding together for white:

This black move is another good try, making miai of A and B:

But white can falsify the D2 eye:


(1 could also be at B4 I think. 1 at B5 or C4 would not work, since then black gets C3 in sente)

White must take care not to get too crazy with the second move. Connections along the side are generally stronger in capture go than regular go (since some of the normal disconnection sequences rely on sacrificing stones), but this white 2 is swallowed up by 3:

This A3 on the other hand might actually be working:


(the variations after B3-A2 get a bit messy, so I’m not sure)

Preliminary conclusion: 2-3 dies. Please let me know if I overlooked something!

Stay tuned for some 2-2 variations tomorrow :slight_smile:

2 Likes
2-2 variations and conclusion

When black starts at the 2-2 point, white needs to aggressively deny the eye-shape with the second move to stand a chance. A soft white response like 3-3 fails completely, since after 3 black already has enough guaranteed eye-shape (the triangled points) to live:

So let’s try this B3 instead for white:


I think black can live with either A or B. A is maybe the easiest, but a bit boring, so let’s look at B instead.

These moves seem natural:

White goes for another desperate-looking kill:

But black has enough forcing moves to set up this net:


(There is no real reason not to play the E2 atari before 13. I just wanted to be fancy.)

One should also consider these attempts for white:


I’ll leave it to the reader to verify that neither of these can kill.

Conclusion: 2-2 lives!

The natural next step will be to check if 2-2 also lives in a 5x5 area.

4 Likes

Problem 25

image

  1. What is the status of each black group?
  2. Can black (playing first, as usual) win the game?

The first part should be pretty easy to read without putting stones on the board, for the second part feel free to click on the image and use the interactive board :slight_smile:

1 Like

A little off-topic, but here is William Cobb’s four-introduction to Go through capture Go, or – as it’s called on the BGA site – Atari-Go.

3 Likes

There is a well-known seki with four spaces:

Summary

image

1 Like

This is seki in regular go, but not in capture go, since black can simply capture two stones! :slight_smile:

5 Likes

Today I randomly got curious about visualizing the winning moves for capture go on a 1-dimensional board, so I dug up my old solver and made this image:


Here the n:th row is a board with n intersections (drawn as circles for visual clarity). The black circles mark the winning moves from the empty board. The final row is n = 24.

It’s easy to see that for any odd-sized board (n > 1), “tengen” is a win, while playing right next to tengen is a loss. (Exercise for the reader: why?)

Harder question: Is it true that for all n > 7, moving at coordinate 2 is a loss, while coordinate 3 is a win? I think this is probably true, and likely not very hard to prove, but I didn’t prove it myself yet.

In the last five rows of the figure, the only losing moves are coordinates 1, 2 and the one adjacent to tengen on odd boards. Does this pattern continue indefinitely, or will there be some more interesting stuff happening?

It could be a fun challenge to solve larger boards. I think with the right optimizations even a board of size 100 might be doable!

5 Likes