Difficulty: Advanced   Keywords: MiddleGame, Strategy, Question

Alex Weldon: Okay... so this is an abstraction of a position I had in a KGS game today.

Vulnerable hoshi: which approach?  

Obviously, the position in the lower center was not quite this neat, but this is basically what it boiled down to. The Black group is alive, and there is another small, squished, but alive Black group in the lower left... they are separated by a string of White stones (linked up to a live group, so they're in no danger). So, essentially, the bottom and lower left corner are played out, but the rest of the board is wide open. There's a White hoshi in the top left.

I'm assuming that, since White has developped a wall facing Black's marked hoshi in the course of attacking the Black lower side group, the most urgent thing for White to do is approach it somehow. But how? I considered all of a through g, but none seemed right.

Now, in retrospect, I'm thinking b or c, but in the actual game I tried something fancy that didn't quite work as planned.

Is this as complicated an issue as I think, or am I overlooking the one, right move?

SnotNose: The White approach has got to be from the direction of b or c, the idea being "use your thickness to attack" or "push your opponent toward your thickness." White a would hardly be worth anything. I think White wants to play c, aiming to uproot Black's corner.

Vulnerable hoshi: which approach?  

SnotNose: Isn't this making territory while attacking? White a later would force Black out toward the center.

Bill: B2. Do not approach strong stones.

Dieter: There is a scent of joseki approach here, while a proverb like play away from thickness is more likely to bring good ideas, which may coincide with a joseki. Applying the principle here, ...

Vulnerable hoshi: which approach?  

B2 plays away from White's thickness. W3 likewise. W3 also threatens an invasion at a (maybe W3 would be better at b or c for that purpose). B4 defends against that invasion. I sincerely believe B4 is better than d, because although bigger, it will leave room for strong white endgame.

Hence, ...

Isolate Black's stones  

... one must ponder W1. Black can also react with B2 at W3. I have no opinion on which result is most desirable, but my intuition is such that I consider W1 and a before the approach at W3, because they go with the principle to attack the corner stone, using the thickness at the bottom, more than W3 does. Again, I'm not saying that W3 is worse, in fact, given the openness of the right side, it may be the correct move.

Isolate Black's stones  

Velobici: W1 should be played to isolate the lower right Black stone from the rest of the board. The high approach is a more severe play to accomplish this goal than the low approach at c. If White plays W1 at c, this allows Black to try and force his way into the center and force White into a low position on the right by attaching at the point that should have been occupied by W1.
White should play W1 to force this Black stone to live locally. In doing so, White should establish thickness up the right facing the lone Black stone in the upper right. That lone Black stone now faces both the thickness just created and a White stone in the upper left (per the original diagram description). White can look forward to an easy game playing against that single isolated Black stone. Black will be limited to about 25 points in the lower right, a huge corner, but not enough to compensate for the isolation of the top right corner stone. That's a 10d take on strategy from a 10k tactician ;)
No follow up moves provided. I dont know enough 44PointHighApproach patterns.

Alex Weldon: Good commentary from all of you, and more or less what I thought. For the record, what I played was the very low approach e, which might be thought of as hamete in this position. Black responded correctly, though, so he lived in the corner, while the position I got on the right was lower than it would have been if I'd played honestly. However, he then tried too hard to press me low and get out to the middle, which started a battle that ended up with me cutting him off and (much later) killing him in the middle.

SnotNose: Agree with Alex that comments are good and interesting. They point to a general question I've had for months now: how does one decide whether to approach a stone (whether in the corner, or pincering an approach) on the 3-rd line or 4-th (here we've even considered the 2-nd line too!!!). I appreciate that this question is nearly unanswerable in general. What is interesting is that, even in a specific case (above), there is room for disagreement. That is, 3-rd, 4-th (and maybe 2-nd) line approaches seem all seem playable, at least for shodan (or higher?) level players. So, a more refined question is: are there circumstances where 3-rd vs 4-th (vs. 2-nd) is only a matter of taste, each leading to different strategies and games but all okay? How does one determine whether the height of the approach matters and when it is just style?

Velobici: When pincering, either one is trying to extert the maximum of pressure is exterted by playing on the same line as the pincered stone. This depends directly upon whether or not the pincered stone is on the same line as the corner stone.
If all three stones (pincered/approach, pincer) are on the same line and one develops from the pincered stone by touching either the corner stone or the pincer stone, that touch will have to be a knights move and is very likely to be touching a weak stone (in contravention of the proverb Don't touch weak stones). The knights move has an inherent weakness that it can be cut. Successfully splitting between the pincered stone and the knights move stone has a bad result for the pincered there are two weak groups to look after.
If the corner and the approach move are on different lines, then the approach can be developed by playing against/toward the corner stone, even though it is weak and will be strengthened as a result. This is often a way of building power/thickness which can then be turned against the pincering stone. That's one reason why a combination pincer/extension from the other corner is such a valuable move. Another reason is that it works toward securing territory while limiting the development of the pincered stone. More often than not, the pincer is also a sente move, so one is working towards two valuable goals in sente as well%%% Whole Board Thinking in Joseki is all about these considerations and what patterns can emerge from different approach and pincer responses.

@ Velobici--Let me re-emphasize my point. Knowing lots of joseki and the general concepts you suggest is not enough to suggest the height of the approach/pincer in all cases. This is certainly true at the 1-3 dan level (this page might be a case in point--though I am skeptical of any move other than c). I cannot prove but I am willing to believe it is also true at the pro level. That is, there are some circumstances where 4-th or 3-rd line are both playable, the difference is a matter of taste. (A database search would confirm this but we all know it is true.) The question is, when is it a matter of taste and when is one clearly supperior? Follow-up question is: are their good heuristics that can help amateur dan (or kyu) level players make a decision without knowing all the joseki? I think this is a hard problem that cannot be solved in a few lines on SL. At the very least, it deserves a small book (e.g., of the length of MonkeyJumpWorkshop). It might require a larger book. Or, it might be too context dependent to approach at all in any systematic way. I am willing to believe this last possibility is true, but I do not know.

SnotNose: I came across a problem in 501OpeningProblems last night that is identical in spirit to this BQM, though it differs in (unimportant) details (exact position of stones, orientation). It is number 270, page 135. In the problem, it is Black to play and defend, rather than White to attack. The answer is the equivalent to c in the diagram at the top of this page (small knight's enclosure). The answer says "If White were to play here, the Black stones would be in trouble because of White's thickness." This is strong support for c. I would be surprised if 9 out of 10 dan level players didn't play c without much thought. Sure, one can ponder other approaches, but it is hard to imagine one's hand not flying to c in an actual game. It is also hard to criticize this move, in this context. What happens after White c is a separate question. Some new ideas occurred to me last night.

After c  

SnotNose: Black is very likely to play B1 or a after White plays white+circle. In either case, W2 is still sente against the lower right corner, so Black will need to make another move. White now has good influence facing the top and can extend perhaps as far as b for the beginning of a moyo.

After c  

SnotNose: B1 is a viable alternative for Black. Here is the normal continuation, which gives a similar result to the previous diagram (Black life, White thickness and the start of a moyo). White can be satisfied here. But, I wonder about W2 at B5?

After c  

SnotNose: Black can play a and then move out toward the center. Black's stones would be without a base and come under attack. Or, Black can play b.

After c  

SnotNose: If Black now plays a, we get a similar result as above (White thickness & developing moyo. If Black plays b, White will get the corner and Black might have some prospects on the right side. I think this (Black b, White a) would be the worst result for White.

After c  

SnotNose: Black has follow-up moves at a or b. White has made an investment too large for the profit and Black dodged the attack. So, I conclude that the W2 of the previous diagram is bad and White should stick with the more standard continuation.

dnerra: I have some doubts this sequence. I know the word "underplay" doesn't exist, but W2 and W4 look like one should create it. Can't White play

W2 at B3? This is exactly the kind of fight I would appreciate with such a strong wall nearby. Then, I doubt Black could find any ko threat to fight the ko if White ataries at c with W4. After Black connects, White could greedily connect at a or extend to d. The corner becomes much larger than in the given variation, and Black has about one and a half eye less.

SnotNose: All in all, it is hard to see how White goes wrong after c. Most likely, she will get thickness and a moyo, while Black lives in the corner in gote.

Charles My idea.

Contact play  

In such positions it can be good to play at W1. The reason is that Black a, White b, we have a 3-3 invasion position, in which Black has blocked on what is definitely the wrong side.

SnotNose: Isn't this SelfIndulgentReading? No way Black plays a. A hane on top of W1 or b seems more likely, no? (I'd play the former.) Black wants sabaki here (a live Black group so near White's wall would be good for Black, unless White gets outside influence of equal value). Black a is way too heavy in this context. That doesn't mean W1 is wrong. But Black a sure seems too bad to count on.

Andrew Grant: But the hanes look bad as well. White will most likely crosscut and start a fight. With White's wall so close, it's hard to see Black doing well. Fighting in front of your opponent's strength is dreadful.

SnotNose: a has been argued to be bad. I'm not convinced hane is bad (showing some variations would be helpful). There is also Black d. Why would Black play a heavy move like a in the presence of White thickness? I think a lighter, sabaki-seeking move (like c, b, or maybe d) might be better.

Contact play  

SnotNose: For what it's worth, here's a hane and cross-cut example. Is this bad for Black? Or, more to the point, worse than the descent below black+circle? Maybe White wouldn't play out of atari at W5, but cut at a instead, to set up a ko? (Note: considered B8 at b but the peep of W8 looks too severe.) The aji of black+circle, making Black c sente, will be helpful to Black in the coming attack.

Bill: The problem with B4 at a is that White has a ladder breaker in the top left corner.


Andrew Grant: White will play 1 and 3, leaving the marked stone in a silly place. If 2 at a, White will be happy to fight the ko as he already has a profit on the lower edge.

dnerra: This is a disaster for Black. Maybe you are right that black cannot play the ko (I'd guess he has to), but then B8 in the previous diagram has been a very bad move.

Bill: Yes, instead of B8 the solid connection looks better.

SnotNose: This is better for White than if Black had played the descent? (This is the question we're addressing. And what about the jump to d suggested above? Question is: is the descent Charles suggested the best Black can do? It is so obviously bad and heavy, I would hesitate to play it. I'd have to be really convinced something lighter is not better here.) Ok, change the marked stone to something else.

Contact play  

SnotNose: Torn between this B8, a or b.

Andrew Grant: Difficult, but I think I'd play a now as White.

Contact play  

Black 2 and 4 aren't as good as they might look, as a and b are miai for White.

SnotNose: So, perhaps, Black protects b before playing B2, leaving the B2-W5 for later. If White is so kind as to protect this, Black has sente to extend up the right side. If White prevents the extension, Black has the B2-W5 reduction (in gote.

Anyway, we could go on with this for a while. I'm just as unlikely to play the heavy descent move (circled point) after White's attachment (white+circle) as before. Is it really the one a 3 dan would play? If so, would the 3 dan have thought through all the other, light, moves or is the descent obviously the only move (and if that, why!?!). The resulting wall Black would get seems so terribly useless and Black's top right 4-4 stone has little hope for development. Maybe tenuki is better than the descent?

Black resists  

Bill: After the crosscut, I think that Black can put up a strong resistance with B4 - B6. Don't ask me what best play is after that, however! ;-) I would be quite willing to give up the corner if I could put enough pressure on W1 and W5.

Charles Well, I was trying to illustrate another kind of idea. This thought is certainly sometimes good when White is strong on the lower side. I don't know whether it is best, here.

A couple of points:

  1. I think strong players do try this method of 'make the most obvious reply bad' - just to knock the opponent off course.
  2. The cross-cut fights aren't easy to read out, but at least there is some reading to be done; suggests White's position is alive.

Something else to consider.

Contact play  

Suppose this happens. Are the white+square stones still working for White?

I think a white invasion around the circled point would show that the white+square influence hasn't been completely neutralised, even though White played for territory in the corner.

SnotNose: Thanks Charles. The admission of some degree of uncertainty/ambiguity is what I was after. (See, not so clear cut afterall!) The suggestion of W1 is interesting and whether it is best here we can debate for longer than we likely care to. Whether the white+square stones are getting 100% use in the above diagram is also worthy of debate (above my head I think).

To sum up, if I may, the knight's move (my suggestion) is the simple path to a satisfactory result. Charles' attachment idea is one way to complexity, which is sometimes the only way to victory (or is otherwise advantageous given relative strengths of players). Black's best reply to the attachment is likely not to initiate the out-of-order wrong-way-block to the 3-3 invasion joseki--but I'm not sure this has been demonstrated. (I'm waiting to be convinced otherwise.)

All in all, a very interesting BQM debate.

Black resists  

Dieter: Isn't this the best way for Black to treat the crosscut ?

Velobici: The diagrams have dealth with low approach (c), inside attachment (f) and 3-3 invasion (g). We dont have a single diagram here showing a continuation based upon b. Have we discussed and ruled out the high approach (b in the top most diagram)? If so, could someone explain why b has been ruled out? (Looking for some dan players to humor a 10 kyu here and help him improve. Dont worry I wont ask about a, d, and e. ;)

SnotNose: I'll take a stab. I'm not 100% confident in my thinking but here goes...b is a move for outside influence. The same objective can be obtained better with c, as shown in the diagrams above. Compare the following two diagrams.


SnotNose: White has a large gap between W1 and white+circle, marked with circles and no real gap at the edge (makred with a square). (If W3 one space below a then there is a huge gap at the edge--too big.) Because of the large gap between W1 and white+circle, Black has more scope for reduction/invasion of the moyo on the right. Basically, W1 and white+circle don't work so well together and, therefore, W5 is too big an extension.


SnotNose: This way, the gap between W3 and white+circle is smaller, but the gap at the edge is bigger. However, White wants to emphasize the center (that is where her strength is), so it is better to have less of a gap in the center at the expense of a slighly bigger one at the edge. Moreover, I think White a is sente, so the gap at the edge can be eliminated. White has a better wall for her moyo this way and can safely play W5.

SnotNose: A general point: I usually consider the most standard approach move first (c). If that seems to work well, I might think about other moves, but they have to really be clear improvements for me to play them. Circumstances where I'd play b are when a Black cap of c (a Black play two spaces to the left of c) is too good for Black (Black would have to have good prospects on the bottom for this case) or when I do not want to face a Black pincer. Neither situation applies here.

I consider Charles' attachment idea typically when taking territory in the corner or outside are miai (usually this is a move I consider for invasions of moyos). This isn't so clearly the case here either, though it does still seem playable.

On balance, I strongly favor the most common move (c) unless circumstances strongly dictate otherwise. Playing a less common move when circumstances don't warrant it invites trouble. That is, there is a larger probability I will regret not having played the most common move.

So, the most common move is not always best but there is a reason it is most common. It is quite frequently best and one should have good reason to play differently.

Put another way, I'm less inclined to feel I have to justify c over b than the other way around. Someone would have to convince me b is better to knock the idea of c out of my head. Showing one or two favorable variations is not enough to convince. It really has to be a very good argument, typically based on more than just exploration of variations. The reason is that I cannot read all variations with every move (who can?) so need to base thinking on higher level concepts. Reading is only part of the answer. Ultimately one has to support moves with some other idea or else it is not possible to remember when they are good or bad.

Another point of view


Low approach  

This seems the ordinary idea for me after W1.

Here B2 and B4 do something definite, in sente.

High approach  

If Black plays this way after W1, it is perhaps slightly better for White (?), than before.

Kitani style?  

Bill: Black can secure the corner with B4 - B6, a la Kitani Minoru. Then if W7, B8. [1]

Black forces  

As for who gets sente here on the edge, I'm not clear. B1 and B3 will force a reply, in most cases. White would become thicker outside.

dnerra: Side question: Isn't the usual shape for B3 to play on the third line? Charles I wondered about that too. It would be 'more' sente, I guess. And good idea here, therefore.

White forces?  

Here after W3, Black can consider tenuki. The aji of White a is quite serious, though.

unkx80: I thought the aji at b is even worse. =)

Charles Ah yes - more common in pro games, I find, though both are seen. Interesting, but would need its own page.

dnerra In this specific situation, it seems to me that b is very severe. I would prefer to defend in gote with black (which would mean playing B2 at c btw).

Charles Having looked at examples, I agree with that.

[1] SnotNose: Never saw this before so I'm just checking the safety of the corner in the next few diagrams. I'm not drawing any conclusions yet (needs more thought). Feel free to point out obvious stupid moves or other ideas.

Corner safe 1?  

SnotNose: Is ko possible?

Corner safe 2?  

SnotNose: This W9 doesn't help. White a, Black b.

Corner safe 3?  

With B1 Black can live without ko. B7 might simply be at B9.


SnotNose: Awesome! Good stuff Bill. The cut at a is big. Is this enough for White? If not, then it would appear that White's wall goes to waste.

Perhaps provoking a fight with an attachment (Charles' idea) is better than an approach. (I gotta say, it will still be hard for me to keep my hand from flying to an approach in an actual game. Since it is so unlikely players at my strength would know Kitani's defense here, that's probably okay. We're way up the dan ladder at this point.)

BQM131 last edited by on January 26, 2015 - 07:58
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