The situation is this: your opponent is between two groups of you: one groups of yours is weak, one is strong. Is there a general strategy to attack from your weak side or is this non-liquet and dependable too much from specifics to make a general rule?

I believe there is a general principle like this. Maybe it is even more general. You should help your weakest stones. If two white groups are fighting with a black, group then it is often crucial to help the weaker group. If you can attack, then all the better.

Of course it may not always be clear but for a general principle then I think it is one of the more dependable that you should play from your weakest group. One could try to qualify by adding some condition that it is not throw away stones, but I think it detracts from the message

Anyway, this is from the top of my head. Probably there are some more specific principles that don't occur a as readily to myself.

The general principle in the Japanese proverb tells BOTH sides NOT to go near thickness. This is usually expressed, not entirely felicitously, as “Stay away from thickness.” But the Japanese is a little lax, too,

kvasir and John suggest what is usually right but exceptions occur.

If someone knows an exception I'd be interested.

It can be good to play and live in front of the opponent's strong wall if any disadvantages from being attacked are small.

I mean, are there exceptions to the initial statement (attack from the weakest group)?

Attacks can have different directions or timings, for which the currently weakest group is an important aspect but not the only one. E.g., it might be sacrificed as a sample cause for an exception.

Imagine a situation where White has a strong group (thickness) on the right side of the board, facing the left side, and also has a single stone on the left side at C4. The lower side is empty.

Imagine two scenarios. (1) Where should Black play? That's not immediately relevant here but is relevant if you are prepared to think about it for a while. (2) Black chooses E3. We now have direct relevance. Where does White play? D3 or G3? No-brainer?

The real problem is to decide what level of player you are dealing with. The usual proverbs (don't make territory in front of thickness, stay away from thickness, etc) are aimed at weaker players. The evaluations are crude: simply territory-based.

But as soon as you try to talk about such positions in a more advanced way, it's not so much a case of whether the beginners' heuristics work as, rather, how you evaluate the result. One common example is, as Robert mentioned, overconcentrating the opponent's thickness could be good. But that kind of play is a sub-set of a bigger evaluation technique which is really only feasible at pro level: amarigatachi. A further problem with that is that you see it only rarely in pro-pro even games (because they generally have good technique/suji). You will see it a fair bit lot in old Edo low-handicap games (look at Shuwa if it interests you) or in modern pro-am high-handicap games. You will nowadays, though, see it in AI-pro games, not often but enough times to be interesting. But amashi and amarigatachi take too much time for most of us to study properly.

The rule doesn’t apply to your scenario because one stone is always negligible. What the rule has in mind is that you have some strong structure on the right side and on the left side two or three or four weaker stones in a loose formation and the opponent splits you with a stone in the middle then what to do and from where? If my proverb is right then you should approach with a stone that naturally extends your weak group, so from the left side.

Your OP didn't state that, and in the revised version I'd be a bit worried if I'd spent as many as three or four "weak" stones that had to be worried about a lone invader. But trying to envisage what is now described, I'd also want to know why a cap isn't being entertained?

This discussion would be more interesting with diagrams Not that it is easy to come up with a diagram in this day and age when someone may wish to refute it with computer analysis but I don't require a diagram for a computer brain. I just miss something to use as a guide for the disposition of forces that is described.

Go proverbs are not iron clad rules, they just give easily understood suggestions. For many proverbs there are positions for which following the proverb works out badly. For that reason discussing a specific position is a good idea. As John Fairbairn mentioned, in a situation described by Pippen. depending on the specific position playing a capping move might be good, or even tenuki.

If we are looking for a general rule, the general rule should be to utilize an attack to achieve the biggest benefit. If insufficiently significant benefit can be made, hold off on making the attack.

It is difficult to get more specific than that while maintaining accuracy.

Reasons for attack could be:
- Strengthen your own weak group
- Kill the opponent
- Keep the opponent weak so that points can be made efficiently elsewhere
- Make points directly
- Create a situation where opponent's moves are less valuable than yours
- Build strength to prepare for another objective (e.g. an invasion)
...

There may be various benefits to be achieved from attacking stones, and depending on the board position, you can utilize the attack to achieve some purpose.

I think that type of model for thinking is more useful than a general proverb about extensions.

_________________
be immersed

Here's an answer from the Oracle - katago versus Nakane Naoyuki 9-dan in his quest to see how far he can go in his 12-game uchikomi match, changing handicap at each loss. This is Game 3, on 2-stones. He lost and so has to try 3-stones next.

The diagram below shows the moves up to 35. But step through to move 16 then guess White 17.

Do you know where one can obtain the game record? With some searching online I found some videos of earlier matches at https://www.youtube.com/@nakaneigo-doujyo/videos but seemingly not of the match you presented the opening of here.

A KataGo review of my latest game reminded me of this thread. First position:

Click Here To Show Diagram Code

[go]$$ Position at move 48
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . O . . . . |
$$ | . . . . . . . . . . . . X O . O . O . |
$$ | . . . . O . . O . . . . X X O O O X . |
$$ | . . X , . . . . . , . . . a X X X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . O O X . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . O O O O . . . , . . O O O O O . . |
$$ | . . X X X . . . O . X X O X X . . . . |
$$ | . . . . . . X . . . . . X . X . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

Although KataGo's focus is on the cut at A in the upper right, my focus - and I assume many of yours - was on the upper left. The thread is a giveaway but where would you play as White? And even with the hint KataGo's preference may still surprise you.



Position 2:

Click Here To Show Diagram Code

[go]$$ Position at move 53
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . O . . . . |
$$ | . . . . . . . . . . . . X O . O . O . |
$$ | . . . X O . . O . . . . X X O O O X . |
$$ | . . X X O . . . . , . . . . X X X . . |
$$ | . O . . . . . . . . . . . . . . . . . |
$$ | . O X X . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . O O X . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . O O O O . . . , . . O O O O O . . |
$$ | . . X X X . . . O . X X O X X . . . . |
$$ | . . . . . . X . . . . . X . X . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

A few moves later, another example fitting this thread.

It is an interesting example. Looking at the second diagram, I didn't like the cramped ugly shape Black has ended up with.

If I were playing Black, I'd like to challenge W's use of the new AI orthodoxy of leaving the wall undefended - by playing 49 at D9.

Playing your 48 at C12 is ambitious AI style to gain maximum points, but I'd personally worry I'd have to be KataGo strength to manage the complex fighting in any counter-attack.

[Edited: to remove my confusion over move numbers and my first unsuccessful attempt at using a diagram]

Probably the reasons of Katago's suggestions are:

• it's the common base of two weak groups;
• the move is reverse sente.

lightvector:

(;SZ[19]FF[3]
PW[Katago]
PB[Nakane Naoyuki]
HA[2]
RE[W+R]
US[GoGoD95]
AB[dp][pd]
;W[pp];B[cd];W[cn];B[fq];W[ed];B[ec];W[fc];B[dc];W[fd];B[cf];W[nc];B[jd]
;W[lc];B[oc];W[nd];B[pf];W[jf];B[kc];W[le];B[hd];W[hf];B[ge];W[dd];B[cc]
;W[df];B[gf];W[cg];B[bf];W[dg];B[hg];W[dj];B[if];W[ig];B[he];W[qh];B[re]
;W[je];B[ld];W[ih];B[lb];W[ob];B[pb];W[mb];B[gi];W[md];B[mc];W[og];B[qn]
;W[qo];B[pn];W[np];B[pk];W[gp];B[gq];W[ip];B[of];W[ng];B[ke];W[kf];B[dk]
;W[ej];B[kd];W[mf];B[lg];W[lf];B[dm];W[eo];B[cp];W[dn];B[em];W[hi];B[gj]
;W[gh];B[hb];W[gb];B[jb];W[ck];B[cl];W[bk];B[hj];W[hh];B[ek];W[bl];B[gn]
;W[hl];B[gl];W[hm];B[hk];W[jl];B[jj];W[gm];B[fm];W[hq];B[fp];W[fo];B[go]
;W[hp];B[cm];W[bm];B[bo];W[bn];B[ll];W[fl];B[fk];W[gk];B[jm];W[kl];B[km]
;W[il];B[lk];W[lm];B[ln];W[mm];B[mn];W[nm];B[in];W[hn];B[oo];W[kk];B[gl]
;W[bq];B[bp];W[gk];B[op];W[lj];B[gl];W[ha];B[gc];W[ib];B[ic];W[gk];B[mk]
;W[gl];B[mj];W[oq];B[li];W[kj];B[mi];W[hc];B[ki];W[ji];B[jh];W[ii];B[jg]
;W[ie];B[hf];W[fh];B[hb];W[od];B[pc];W[hc];B[id];W[pg];B[pi];W[ka];B[hb]
;W[qf];B[qe];W[hc];B[oa];W[nb];B[hb];W[qj];B[pj];W[hc];B[ja];W[rf];B[hb]
;W[qk];B[ql];W[hc];B[la];W[se];B[sd];W[sf];B[rc];W[mp];B[qi];W[ri];B[rh]
;W[sh];B[rj];W[rg];B[jp];W[jq];B[kq];W[kp];B[pq];W[qq];B[pr];W[qr];B[jo]
;W[lq];B[hr];W[ko]
)