My flabber has never been so gasted in go as it has moments ago. For me this outranks even the news of AlphaGo.



This shape comes up in some commentaries I was working on. It was White to play and apparently he was having a hard time choosing his next move.

I goggled (not googled). It's obvious, innit? A or A or A or A. Any letter you like so long as it's A.

We know the cut's nasty as if White pushes up to the centre, this would be the starting point for the famous Kaizen's Great Plastering Manoeuvre (or Kaizen's Adobe if we want to sound "cool"). And clearly tenuki may always be an option in any case. But in the game I'm looking at, White had decided he had to defend.

But various commentators declined even to mention A. Even Go Seigen.

So I looked it up. The various joseki dictionaries I tried also declined to mention it. At this stage my eyes are not just goggling but spinning round and round.

Next step was to try a database search. Slightly relieved to see 8 examples, but irritable bowel syndrome resurfaces when I see the first example was played by an amateur and the rest were highly influenced by nearby stones - and the last example was some 200 years ago. And no examples from China, the home of the tiger's mouth.

Final step was to try dear old Lizzie. The bitch didn't even acknowledge A as a move worth looking at! When I made her look at it, she came back with a "seven percentage-point loss, loser!" smirk.

Now I can well understand that there may be a more efficient local move somewhere (though the commentators couldn't come up with a good alternative). But for such a move as A to be out of the reckoning altogether is just incomprehensible to me. My flabber and gast will forever remain asunder unless some kind reader can come up with a decent explanation.

And by kind reader I also mean all those strong players lurking out there. It's about time you did your bit, too. For my F and G if nothing else.

Every corner sequence is also about competing for the corner territory. If A, then Black D17 gets the corner. Therefore, except for special positional environments, W must compete in the corner at C18 or C17.

Maybe it would be helpful if John posted the whole board situation?

My guess is that it is too slow. White needs has already played one slow move (kosumi at ), so adding another slow move is just not called for. Side note: this joseki more or less disappeared a long time ago (early 20th century?), if I am not mistaken.

Another view to take might be a bit of tewari: if White only had the three stones of the tiger's mouth, would they play that little kosumi now?

_________________
A good system naturally covers all corner cases without further effort.

I experimented with the situation a bit using KataGo (2500 playouts) on AI Sensei. While it definitely wants to play elsewhere, trying some moves locally seems to indicate that there is a local optimum at the top side star point (K16). This seems to be stable enough that e. g. a third-line tsume need not be immediately answered afterwards.

I'd say that this supports the view that the tight tiger's mouth is (usually) too slow at this point (I think it was Kirby who called such a move a »slownection«). Now I'm curious: what did the game look like where a professional player wanted to find a good defensive move there?

_________________
A good system naturally covers all corner cases without further effort.

I attach the complete game. I didn't post a fuller position because I didn't want to put ideas into people's minds.

I liked Robert's reminder of the relevance of the corner, but as he will see from the game neither player saw fit to play there after the first couple of moves. I think we can conclude from that, as I imply in the previous sentence, that the corner is (at most) relevant but not specially important. I claim no insight beyond that - just more bafflement.

I also like the concept of slownection, though I had it down as a knotwilg idea - or was it synergy with Kirby? Either way, a good idea. Slowness did in fact occur to me but I assumed it wouldn't be a detrimental factor because this was a Black move played with no komi by a player known for his tight plays.

Another element I kept hidden initially was that there was another tiger's mouth that is a joseki move and made by no less than Dosaku - but Go Seigen called it a mistake. The move he recommended was the tight connection 32*. The game move allowed Black to settle himself on the lower side, which negated much of White's thickness.

Two tiger-mouth mistakes in one game by two of the greatest players ever - you may understand better now why my eyes were swivelling. Tyger, tyger, neither right/In the corners of the board./What mere mortal eye/Can frame thy sheer perversity?

13 of course was the shape move Dosaku chose instead of the tiger. It looks flimsy, but actually when I ran Lizzie and forced it to concentrate on this area, this is the only move it showed. So kudos to Doteki, surely. Who forced Dosaku to resign anyway.

Doteki is hailed as the greatest prodigy in go. Much of this praise is due to a game he is traditionally assumed to have played when he was 13. We now know it was played several years later, so a little of the lustre has to be rubbed off. But only a little.

* If a tiger is a slownection, what is a tight connection? Dosaku wouldn't have this one on his radar as it wasn't played until 1943.



The numbers guys might like to know that Go rated Black 59 as a five-point loss. I'll leave you to work out the correct move. It's rather hard. I'll post the answer later, though I may need a reminder.

I would never have expected that a pun-term coined by a non-native speaker would find clemency with you, John. "Slo(w)nection" is the main pattern in my ongoing analysis of my 2021 made mistakes with KataGo. Of course someone else, like Kirby, may have come up with the idea or term synchronously, or I have subconsciously borrowed it.

OT: I do think this is such a case. The cut "does not exist" so protecting against it is slow, even with a hanging shape which has supplementary benefits. Sure, there's aji in the cut but this can be dealt with more efficiently by a bigger extension.

That's hindsight: I might have expected the hanging connection too, which would prove that I'm indeed intuitively guilty of "slownections".

Edit: two comparisons

Click Here To Show Diagram Code

[go]$$W
$$------------------------
$$| . . . . . . . . . . .
$$| . . . . . . . . . . .
$$| . . . . . O . . . . .
$$| . . X . . O . a . . .
$$| . . . X . X O . . . .
$$| . . . . . X . . . . .
$$| . . . . . . . . . . .[/go]

Would you play A if there were such an apparent ladder as here, where the supporting stone is one space closer?

Click Here To Show Diagram Code

[go]$$W
$$------------------------
$$| . . . . . . . . . . .
$$| . . . . . . . . . . .
$$| . . . X O O . a . b .
$$| . . . X . X O . . . .
$$| . . . . . X . . . . .
$$| . . . . . . . . . . .[/go]

In this corner pattern, the known move is B, not A. The reason is also efficiency.

I'm sorry, it seems I mixed up the names Kirby and Knotwilg!

_________________
A good system naturally covers all corner cases without further effort.

Wait- isn't the position of the game different than the position of the initial post? The game move is at N6 instead of O6. (Edit: John has corrected the file.)


A slowernection?

Did you actually check with Katago? In my games (15-block) Katago seems to like A about as much as B but often prefers C. I know that A looks slow but it doesn't leave any weaknesses so seems fine.

Some tewari analysis:

Click Here To Show Diagram Code

[go]$$B
$$------------------------
$$| . . . . . . . . . . .
$$| . . . . . . . . . . .
$$| . . . . 2 . . . . . .
$$| . . 1 . . 6 . 4 . . .
$$| . . . 3 . 5 a . . . .
$$| . . . . . b . . . . .
$$| . . . . . . . . . . .[/go]

(Later, or now, White exchanges a for b).

1-2-3 is of course joseki. 4 is not the most normal extension, but I don't think it can be faulted either. 5-6 is generally considered OK by AI. a for b is a slightly negative exchange for White; it's just pushing from behind, it doesn't really have any prophylactic purpose. Black a would not be sente. This isn't an argument that a better move exists for White in the original position, only that she should seek an alternative.

Yes, sorry. There are three versions of the game extant, with this move pair being one of them. This one seems to be a simple typographical mistake and nowadays O6 is considered correct. There is also a doubt about the last move, which some records have at K16. J16 seems preferred now. The third version is simply one where both apparently wrong differences apply. There are also versions with Black 1 in the lower right (i.e. a different orientation), which is a common mark of respect in game collections (respect to Dosaku in this case) but that seems to upset westerners who worship the Rising Sun and see the top right as the source of the sun. Many of those would put a macron on it as well, of they could.

I will change the file.

It was interesting to check that position with and without komi. The usual moves on the left side are not that bad with komi but the move in question was worst of the bunch at -1.1 pts. which is really not so good but O7 seems to be good. Without komi they are mostly pretty bad at around -1.1 to -1.5 pts. except O7 that is at -0.4 pts.

One might get different results with more playouts (and better weights) but it appears that sente to play E16 is strongly preferred in the no-komi game.

Possibly the answer to why "good shape is bad" is that keeping sente is a good way to keep a lead in the opening. In an even game that is not necessarily going to be good enough so the value of sente will be lower and you can see that in local gote moves being closer to the best overall move. This seems counter intuitive but maybe it is really what we are seeing?

I just found another commentary by Takagi Shoichi, which still ignores the tiger's mouth but does give what seems to me to be the best way of looking at this.

)

Takagi regards Doteki's move (as Black) at A as "unusual". He argues in favour of making a force at B instead, then after White C, Black can push up at D. I haven't looked at the details, but I imagine that force negates Kaizen's famous adobe sequence. In other words it is an indirect way of covering the cut. Is this a "fastnextion"?

There was a time when I used to think of shape as something static, an aspect of stones of one color, like "the bamboo joint" or "the table shape". I haven't got rid of it fully yet but at that time already people like Charles Matthews I believe were vehemently arguing against that kind of view and urged us to think of shape more in a dynamic way, at least taking opponent stones into account. A basic example is the difference between a diamond and a ponnuki. A ponnuki is the result of capturing a stone which was there in the first place. It arose. Putting a diamond shape in a random area of the board is not nearly the same.

John is probably wondering why I'm preaching to the choir here but I think the original post has a scent of this "static" approach to shape: how can a tiger's mouth be bad (or rather not being taken into account by pros and AI as a viable move)? Well it's not so much about the tiger's mouth but about its relation to the extra stone, which practically voids the cut already and the space next to it, inviting a more efficient extension;

Click Here To Show Diagram Code

[go]$$B
$$ . . . . . . . . . . . .
$$ . . . . . . . . . . . .
$$ . . . O . O O . . . . .
$$ . . . . a . . . . . . .
$$ . . . b 1 . . . . . . .
$$ . . . . . . X . . . . .
$$ . . . . X . X . . . . .
$$ . . . . . O O O . . . .
$$ . . . . . . . . . . . .[/go]

I'm only slowly getting cured from my table shape addiction. In a recent game (I'll dig up the actual board position later) I played it while KataGo only considered the peep at A and the speedier protection at B.

I remember a similar pattern on some BadukTV commentary. IIRC, the guy was saying that 'b' was efficient in terms of territory. But stronger players sometimes start to play a little closer (one to the left of b), and even to the point of playing 'a'.

Again, IIRC, he was saying that a move like 'a' offers better opportunity for counter attack to the shape. So it's a tradeoff between having efficiency in surrounding territory on top vs. losing some of that efficiency to make a stronger attack on the left later.



Sounds about right. Knotwilg is more creative than me with that kind of stuff. Though, we do both have "K" names.

_________________
be immersed

I was just reviewing a game from 1851 and I thought that a joseki looked familiar and so here I am.

In this game, Black extended at 29 instead of playing the tiger's mouth at 'b.' The Commentator does not discuss 'b' nor the Kaizen sequence. Instead, he states that 25 (preventing White's press) for 26 is now [decades ago] considered unfavorable for Black. Black should just play 'a' instead of 25. Of course in this position Black 'a' is an extension from a wall.

Click Here To Show Diagram Code

[go]$$Wcm24
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X O . . . . . . . . . . . . . . |
$$ | . . X O . O . . O . . . . . X . . . . |
$$ | . . O X O . . . . , . . . . . , . . . |
$$ | . . . X . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . b . . . . . . . . . . . . O . . |
$$ | . . . . 4 6 . . . . . . . . . X . . . |
$$ | . . . 2 3 5 . . . . . . . . . X O . . |
$$ | . . X . . . . . . . . . . . . X O . . |
$$ | . . . , 1 . . . . , . . . . . , O . . |
$$ | . . . O . . . . . a . . . . X . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

The commentator states that when black pushes at 29, the continuation to 37 is natural. So there's that. The commentator does not mention the different variation mentioned by Takagi Shoichi.
To speak more about this joseki, the Commentator mentions that with the ko, Black cannot ignore White 'a' otherwise White will kill with White 'b,' Black 'c,' White 'd.'

Click Here To Show Diagram Code

[go]$$Wcm30
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X O . . . . . . . . . . . . . . |
$$ | . 9 X . 0 O . . O . . . . . X . . . . |
$$ | . . O X O . . . . , . . . . . , . . . |
$$ | . . . X . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . O . . |
$$ | . . 3 1 X X . . . . . . . . . X . . . |
$$ | . a . X O O . . . . . . . . . X O . . |
$$ | . d X . . . . . . . . . . . . X O . . |
$$ | . 8 . 4 O . . . . , . . . . . , O . . |
$$ | . . 2 O 7 . . . . . . . . . X . . . . |
$$ | . b c 6 5 . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

This is Game One of the Nijubanjo between (W) Shusaku 6-dan and (B) Sekiyama Sendaiu. Black ended up resigning. The commentary is from Invincible and was based on a commentary by Ishida Yoshio.

So, it seems like the pushing move is OK. My version of KataGo says that the Tiger's Mouth loses over 2 points. Still, the tiger's mouth seems like a safer move. So I got my take-away.