I have used this title for a special reason. But first try a couple of problems:

(1) Identify the weak groups



(2) Identify which groups are not yet alive



I am referring here to "All About Sonoda's Proverbs", a Japanese book I praised highly here recently. It continues to astound me.

This position comes from his book. Most of the book is concerned with evaluating positions at this sort of stage in the game, and he discusses each by beginning with a phase of "sorting out the conditions" (i.e. identifying the main factors - it means conditions in the sense of prerequisites rather than the ambience). Typically he lists 3, 4 or 5 "conditions" and then shows how these determine the area to play, and then how the move to play (more than one may be possible) follows from that. Often he also shows how moves that look good are inferior because they don't take account of the "conditions".

He will then move on a couple or so moves and repeat the exercise, so we can see how the "conditions" change - I find this perhaps the most fascinating part.

Sorting out the conditions does not follow a strict algorithm, but certain themes constantly appear, and the format is consistent. Just to give a flavour, the position above has the following "conditions":

(1) Areas not alive: White in the lower left, White on the left side, the Black group from the left side to the centre, Black in the upper right
(2) Areas with many stones: Centre-facing by Black and White (this is explained by some hatching on the diagram which I can't reproduce)
(3) Power relationship between Black and White in the centre: roughly equal.

I don't want to discuss the actual moves, but just to avoid too much frustration, the best next move for Black is the knight's move to J9. That move takes account of the conditions. The tempting looking move F17 is inferior because it has no connection with the conditions.

The bit I want to focus on is not in Sonoda's book. It is 'weak groups'. This is one of the most widely heard go terms in English, and a typical conversation might be that the one or both of the White groups on the left is not weak because they are chasing a weak Black group. But the direct equivalent of 'weak group' is rather rare in Japanese. The pros are much more nuanced. Some of the nuances exist in English: a group can be 'thin' or 'heavy' for example. But we still revert and say thin groups or heavy groups are 'weak'. Actually a thin group can be alive but thin because it has straggly bits than can be cut off, but that's a bit abnormal, and I'd say the commonest word in Japanese is not weak (yowai) but thin (usui). But thinness applies mostly to the middle game and it's often the opening we want to evaluate.

Sonoda's solution is to talk about whether a group is alive or not (with shades such as 'not clearly alive'). I don't think he ever uses the word 'weak'(yowai) at all - it's certainly rare. That has astonishing implications. 'Weak' is a trigger word for 'attack'. 'Not alive' is a trigger word for ... what? It encourages thought, diversity. It also focuses the mind on what (and thus where)is important. Look at the top right. You can never say Black is weak up there - he even has a majority of stones. But he is not alive (the 3-3 point is the linchpin).

I think every single evaluation (maybe a few hundred?) begins with an assessment of which groups are alive or not. Sonoda obviously thinks that is the first factor to consider. Although, in this case, J9 is called for, and that can be seen as acting mainly on part (3) of the assessment, the identification of (3) as a factor in an area concerned with a running fight (seriba) stems from factor (1).

This is typical of Sonoda's approach, but let me stress again that he gives considerably more detail.

My goal was to focus attention on the words he uses - hence "Talking Points".

What does Sonoda mean with "not alive" in the cases of W left and B upper right?

This looks like a 'balance of power' example from "Attack & Defence", where J9 holds the balance of power for both Black and White, because both groups are unsettled. Whoever plays there first gets the advantage.

Worrying about the status of w's left side or b's top right looks like a distraction.

Here's another example from for you to try. I found it challenging and you will probably face an extra challenge in that you will not have had the benefit of reading Sonoda's earlier explanations. But still...

Black to play. Find the conditions (three) then choose the area to play, then choose a move.



Hint:


Answer:


A couple of "incidentallys". One is that, while looking for the above example I noticed a "weak group" (yowai ishi) reference. It doesn't alter what I said, but the irony amused me. Second, because I am talking about words and the effects they have, it dawned on me that we usually say "Where does X play?" The Japanese usually say "How does Black play?" I need to think about it more but I suspect there's a lesson there.

You know my standard answer to these questions is: read the book.

But I did exactly the same double-take as you. I was able to shuffle off my doubts almost instantly. For example, over the years I have seen the remark that the 3-3 point is the weak point of the hoshi stone so many times that it actually irritates me. But after this example it dawned on me what that really means. A lovely satori moment. The left side took a bit more thought, but not much.

On past form you will be looking for a classic, watertight definition in the manner of Aristotle. Like very many areas where we use the human brain as the main tool in a very complex world, go is better suited to Wittgenstein's fuzzy families. There are so many examples in this book that by the time you've looked at them all, you will have created a fuzzy family of your own. It will be a little different from mine and we'll both be different from Sonoda and everybody else, but we'll be able to talk to each other in exactly the same way that we can usefully talk about birds and yet still disagree whether a penguin is a bird.

I haven't completed the book so, while I already have a prototype in mind, I am still forming my own fuzzy family in my own way. FWIW a couple of the things that loom large in my mind are that 'not alive' definitely does not mean 'dead' or 'dying' and that (in the same way that thickness is not really thickness unless it can function as thickness) life is not really life unless it functions in the way that was intended for it (which is usually to make significant territory). So if you have a group that lives with the bare two eyes you have actually failed. White is in danger of that happening on the left side (he can be bullied - maybe not now but it's easy to see how that can happen later). In contrast, the Black group in the lower right may have to answer a forcing move or two, but it ends up making the modicum of territory it was designed to make.

One extra point that influences my thinking above is that for 'alive' Sonoda usually uses the form ikite iru. It's possible to regard that also as 'being alive', i.e. being active, living life to the full.

In other words (in the jocular but deep manner of the Tao of Pooh), "Fuzzy-wuzzy was a bear, Fuzzy-wuzzy had no hair, Fuzzy-wuzzy didn't care, Fuzzy-wuzzy wasn't fussy, wuz he?"

Edit: I've come back to this because I said earlier that Sonoda did not outline his conditions in an algorithmic way. That's true but misleading, so I just wanted to stress that he does talk about how to come up with the conditions. On brief:

First, decide what's alive and what's not alive. One basis for this is the often overlooked point that playing close to a live group (his or yours) is incredibly small. That can lop off huge areas of the board where you don't have to rush to play.

Second, be aware of the big points in terms of the number of stones. Various aspects, but e.g. if several groups are not alive, the ones with most stones get attention first. Sounds obvious but the trick there is recognising what's alive and not alive in his sense, and there's also a need to understand what a big point really is - nothing to do with territory counting (there is no such counting at all in his book, remember).

Third, consider height and width (3rd and 4th lines). E.g. if there are several weak groups the 4th line is more important than the 3rd. Width (haba) is a big concept that needs a big treatment for most English speakers, methinks, but I might give a peek at it tomorrow with another Sonoda example.

Very many examples and words with still not more than fuzzy ideas do not help me at all.

What Sonodo appears to mean is instability: Black versus White playing first at a particular group makes a great difference with respect to its connection, life or development potential.

I have defined these concepts in detail elsewhere. I.e., there is no need for remaining fuzzy and sometimes making mistakes because of not knowing exactly what is being meant.

First of all, I regret not being able to read Japanese fluently. Must be fun to read Sonoda's book.



May I try to assist you in cutting the Gordian Knot ?

Dear John, I am afraid that "creating a fuzzy family of your own" might be a bit misleading, as far as "FUZZY" is concerned.

Following your comments on Sonoda's book, I have understood that Sonoda wants to direct the reader's view to several important aspects of a given board position.
The game of Go is extremely complex, so Sonoda's approach is offering a lot of example positions that the reader has to work with on his own.
Sonoda's explanations are more or less "FUZZY", just because he is so clever not to torture the reader (explicitely) with HIS "RULES". He seems to know that there is no rule without exception, and that HE is the only one who knows the exceptions of HIS "rules".

Instead, as you have written, Sonoda expects that -- over time and over many examples -- every reader will have developed THEIR OWN family of "rules" for evaluating a given board position.
C-rules, JF-rules, JR-rules, XY-rules, all will be different from the "original" Sonoda-rules, but what is important is that every reader will have understood THEIR rules (and THEIR exceptions) and will be able to apply these. No matter whether any of these rules will be always appropriate (maybe "= right" in Robert's world).


Dear Robert, I am afraid that you will have to read Sonoda's book on your own to eliminate the "fuzzyness" that you feel so sorry about in the moment.

Probably it might not have become clear to you what I wanted to say, so here comes another try.
During my study of shape issues in one-move life-and-death problems, I needed several passes. All of a sudden -- when busy with larger shapes -- I realized that I forgot to mention an important issue with the smaller ones.
And even analysing the shape issues in three-move life-and-death problems had some kind of retroactive affect on the one-move problems.

Sonoda's "fuzzyness" gives the reader the chance to experience a lot of such "flashes of genius".

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

Marcel, not explaining "not alive" by not explaining "big nearby move" does not help, either.

Let me instead explain: there is at least one big nearby move if the local situation is unstable because there is a big difference WRT connection, life or development potential between Black or White playing there first.

Cassandra, everybody (or every neural net) developing his own "rules" takes very much time (or reinforcement learning iterations) if the "rules" shall lead to strong play in by far the most cases. In my case, for such kinds of go learning aspects, two decades. I learn very much more from the explicit principles found and written down after the two decades of preparatory learning. So much indeed that the latter always override the former. The latter can be learnt in a very short time: as fast as one can read and recall principles.

If Sonoda does not eliminate his own fuzziness then why should reading the book help me for this? Have I not already dissolved it?

No. So another of my standard answers: he's 9-dan, you're not. And by your own admission you appear to have done well over the magic 10,000 hours.

I think you need to appreciate that fuzziness can be powerful. Not much of life and not much of go can be put in tidy little boxes. So when, as we constantly do, we meet a novel situation, we need to devise a way to deal with it. We have to improvise. Evolution seems to show that fuzziness is the best tool we have then.

PS I don't think he does mean instability. Why? Strangely enough, Japanese has a common word for instability. He didn't use it. Weak groups are unstable. I've already expounded at length that he doesn't mean weak groups.

I am not joining meta-discussions about 9p, evolution or non-go invention.

For go decision-making, fuzziness / subconscious guessing / creativity are useful but only as an alternative or optional preliminary step before verification by reasoning, reading and counting. If fuzziness suggests some moves for assisting not alive, unstable or whatever-Sonoda-means groups or regions, one must still verify and choose among candidates.

Robert, please do not underestimate that (in the examples mentioned above) the reader is explicitely shown where to look at and / or what to look for.

Clustering a lot of examplary cases (i.e. developing "rules") will be easier than without these hints.

Please do not also forget that the "rules" developed (e.g. by AlphaGo) might be "better" than those which the author originally had in mind.

Blindfolded following "rules" that were developed by someone else (without knowing the background / development history) will not be as effective.

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

OC, non-100% principles must not be used blindly.

EDIT: Like for every source with selected examples, creating my own "rules" was a bit easier due to the selection. Nevertheless, I needed circa 20 years. Topic-driven selection alone does not enable creation of high percentage "rules".