18k vs 5k (AKA how to descend into chaos all by yourself)

I think it was within my grasp to kill a thing or two.

AI says “BOTTOM RIGHT CORNER, LADY!” early on, but as evidenced by the end, I can’t invade to save my life (pun absolutely).

Any comments welcome.


Not sure if it’s any consolation, but when I look at the bottom right I know instinctively that something should be possible, but I have no clue what and how. Just like you I’d just have tried something, hoping my opponent has even less of a clue.


Ah, I do have one more concrete hint: If you want to invade eventually, don’t play moves like P5 before. That makes your opponent stronger there and makes your invasion more likely to fail.


First of all, that was quite a nice game, with good fighting spirit.

Otherwise, I don’t think the lower corner was easy to invade, but this vital point was much easier to see and would have been a winning move.



At move 16 you make a jump instead of a simple nobi:

The AI doesn’t like this and tells you that you should have simply extended with a nobi at the marked point instead.

Your jump doesn’t really gain anything regarding the east side; and it weakens your wall a lot. Now if white plays the warekomi at the marked point, then your three black stones are going to have a bit of a shortage of liberties, and there’s going to be a triple-cut:


Or if instead of the warekomi White decides to cut at Q15 directly, then because of your jump instead of a simple nobi, you’re not going to be able to capture the cutting stone in a geta:


The geta fails because the three black stones are going to be in atari. If your black stone was at the marked point instead, then the geta would succeed.

As a summary, you should have very simple extended:


Here is a quick rule of thumb to decide whether you should jump or extend in situations like this:

  • If you’re pushing from ahead, simply extends;
  • If you’re pushing from behind, then consider jumping.

Here white is pushing from behind, and you’re pushing from ahead, so don’t jump - just keep pushing ahead.


At move 17 you played K17, which is a very good move, preventing the white stone from making a nice extension on the side, and pushing it against your wall.

At move 33 you played this bad shape:


This is a terrible shape. White’s previous move is a contact move, a “tsuke”. In respond to a tsuke, we almost-always answer with either a hane or a nobi. Here, a hane would have been nice:


After this hane, it’s almost impossible for White to make a good shape. The White group is going to suffer a lot, probably going to be surrounded and live in gote.

However, the move you played offers White the opportunity to give you a bad shape, a “torn tobi”:


This is a very bad shape. Your two marked stones are torn apart by the two white stones. The AI shows a simple variation in which White lives by capturing the two black stones at F17 and G16. This is a missed opportunity for Black, all because of a bad shape. If you can recognise that the torn tobi is a bad shape that must be avoided, then this wouldn’t have happened.

In the game, White only captures G16 and lets you save F17, and so G16 becomes a false eye, so the situation remains bad for White. (The situation was already bad for White to start with, because White has created two weak groups in very close proximity, at M17 and H17.)

In summary: avoid the torn tobi!!


I didn’t particularly like G14, but I was afraid W would come down in the middle and start reducing whatever I was trying to build.

Is there a way I should continue to avoid at least some of that danger?

1 Like

Then the next big mistake is at move 67, when you play at L11:

This is a direction error.

You want to build a territory around G10, I understand. But look at your wall at Q16, and look at White’s stones at M13. Both of these groups are a bit weak. Right now it looks like White’s group is weaker than Black’s, but weakness is relative: the stronger the black wall will be, the weaker the white group will feel; and conversely, the stronger the white group will be, the weaker the black wall will feel.

The black stones at K13, however, are already very strong.

When you have a weak group (Q16) and a strong group (K13), it’s usually better to play from the weak group, than from the strong group.

This is why the AI suggests playing at N12 instead of L11 for Black. Playing from your weaker group and making sure that your stones are stronger than the white stones is more important than protecting your territory in the centre.

In the game, White doesn’t exploit your weakness, very much, and creates a second weak group by playing at K9 at move 72. Around moves 85-101, White makes a huge blunder that results in 7 dead white stones on the south side. After move 104, the game looks pretty much over. All the black groups are pretty strong, and White has two super-weak groups which are completely surrounded:

If you just play at the marked point now, the Black groups will be completely connected, and the two White groups will be completely surrounded. White will have to try to make life for the five stones on the north side. White will probably manage to live, but gote. Then you can add a move to make sure the five white stones in the centre, around K9, remain dead.

In the game, White manages to make life for both group. Jlt already pointed out that you could have killed the white group on the north side if you played the vital point N14 instead of M18 at move 145. Indeed, M18 looks like a “safe” move that connects your groups, but your groups are already completely connected and completely safe at this point, so there is no danger at all for you, you can play the vital point N14 and kill the white group without taking any risks.


Yes. Just play the hane at F15. White’s shape is so bad that there is no risk for you.


What is White going to do after the hane at F15?


This is a notorious bad shape that is rarely good, since you are playing a one space jump that is already cut:

A more possible result would have been this and as you can see, black would get two eyes immediately, ending the fight and any subsequent attack at his group:

On the bottom right side, since you opted for this fight, you should have connected. Sometimes the empty triangle is indeed the only move which can connect two stones that are diagonal to each other. After that you’d have sente for a lot of moves in the image:

K9 is a direct threat to all this moyo and it is exactly why you are building that wall. Walls are not meant to be defended, but they are not meant to be abandoned either. They create a sector line (in red), that is used to attack the stones that try to cross it. The usual approach is to expand one of your two sides of your wall to envelop that white stone into the sector line (blue lines, depending on the choice you’d make), thus attacking the stone, thus making the opponent run, thus gaining points elsewhere while the opponent runs for dear life :slight_smile:

Here is a possibility for the invasion. Move 2 is the only one that is not really sente, but given the shape, it is the most likely move to be played in our level. After that move 3 is dangerous and if White drops at 4 to defend, the rest is very possible.

There are refutations of course, but White barely survives and it is still a large gain for Black:

1 Like

G14? I thought that somehow W would start to come south and bring havoc…
(that’s projecting my weaknesses, no doubt about that!)

If White plays G14, you can cut immediately. White is not going to be able to save everything:



What were you trying to build?

1 Like

I think that was precisely the mindset that I was missing. If W was trying to do stuff to the area I was afraid they would, I should immediately search for the “but where are they not playing while they are playing there?”…

1 Like

Some territory around F11 or thereabouts…

This was not the right moment to build anything. At this stage of the game, you have to examine weaknesses. Is one of your groups in danger? If so, defend it. Is one of your opponent’s group in danger? If so, put pressure on it.

1 Like

At move 151, you played P19:

This is a completely-gote endgame move, worth exactly 2 points. It looks like you played there because White’s last move was R18, so you “responded” in the local area.

But White’s move R18 is gote. You don’t need to respond. You should tenuki. Playing P19 now is only worth 2 points; it doesn’t bring you anything else. Your groups are already completely alive, and White’s group is also completely alive, so there is absolutely no urgency to play in this area.

So now is time to take the initiative and play some big moves. There are still at least three areas in which moves are worth a lot more than 2 points. In fact, most areas on the board are worth a lot more than 2 points:

As a summary: when your opponent plays a move which is completely gote, don’t immediately respond in the same local area. Take a deep breath, look at the whole board, and take the initiative.

A good hint that a move is gote is if all groups in the area are already completely alive.


And finally I arrive at this.

The reason the AI believes invading is so easy is because of this weakness of White:


Normally when Black plays a move like this, a second-line clamp, then White has to choose between cutting on the first line at T8, or connecting on the third-line, at R8 or R7.

Connecting on the third-line is admitting that there was a huge weakness, and letting Black get a huge endgame sequence:


Cutting on the first-line means White is ready for a fight. However, in this situation, White would completely break down:


So, cutting on the first-line doesn’t work for White. White can still choose the safe, peaceful option of connection on the third line.

Or… Maybe not?

Because the second-line clamp at S7 was one weakness in White’s territory, but White has another weakness:


If Black plays at A or B, then White must protect with a move at one of the marked points.

So, there are two weaknesses in White’s territory. Black’s best bet is to try to use both weaknesses at the same time, something like this:


Now Black is alive by miai: either connect at A or live in the corner with B.

Now, this sequence that I’ve shown is pretty simple, and maybe not the best. In particular, after Black plays the clamp at S7, White might still have peaceful options by sacrificing the white stone S8 and trying to save some territory in the corner.

For this reason, the AI suggests being a lot more bold and invading directly in the corner:


The idea remains the same: Black should exploit White’s weaknesses. Here Black is threatening three things simultaneously: making life in the corner directly, or connecting to the Black stones at O5, or connecting with the black stones at S9. Because of this three simultaneous threats, Black’s invasion should be successful.

However, in the game, you waited much longer before the invasion, and you strengthened White before you invaded:


Now you can’t connect to the black stones at O5, and the white stones are completely connected. So the invasion is much, much harder at this point.


To avoid that messy thing after g14, i think best way to avoid it is to not play g16 as follow-up on shoulder hit. Push+hane is more common way for building yourself influence while keeping your opponents group small

Bit later on move 55 when you managed to get sente from that mess and successfully got a wall between whites h17 group from m17 stones, you never attacked those m17 stones >__>
Instead c6 maybe keep the pressure on white there and maybe you can build something while chacing them

Imagine what could few peeps make to that shape ^___^


A few more details on that concept then:

Once you extend your wall, usually your opponent will have to run to get out of the sector line. In some cases like this one where the area is wide there is a possibility that they will stay in and fight like in the following image with 2, for example (which was later played in the actual game), but once Black plays 3, these two White stones are in obvious distress.
No eyes, no space, almost fully surrounded, nowhere to run to. Not something good, so White instead of 2, probably runs towards the blue arrow, to head out of the sector line.

Again if after the first jump at 2, White thinks that now there is more room for fighting and responds with 4 to the sector line being extended again, then this is a possible result where White keeps most of the points in the area and gets influence towards the right side. White is out and can connect with either side or make a base, but White didn’t gain much and the cost seems to not be worth the effort.

So, the more usual case is this. Jumping out again and again for a few moves, and Black moving the sector line again and again. White only has one group, while Black has TWO sides and cannot extend both, so this means that eventually the reducing stones of White will outpace the sector expansion, but while this is happening, Black is gaining a lot on the right side or elsewhere (depending on the game).

Indeed, now White is out and it is Black’s turn. (11 and 12 are sente, but not very large moves at the moment, I just put them on the board to complete the area). Black can play A immediately to keep all that area and then try for the right side OR Black can play at B and jump on the right side and later defend at A:

Either way, as you can see, the areas are roughly the same. Even if Black was to be reduced in the center, Black would have been handsomely compensated elsewhere.

Stronger players will have more to add to this, but I hope this helps.

1 Like