It is currently Thu Mar 28, 2024 12:34 pm

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 27 posts ]  Go to page Previous  1, 2
Author Message
Offline
 Post subject: Re: X marks the spot
Post #21 Posted: Thu Jul 23, 2020 3:19 pm 
Dies with sente

Posts: 83
Liked others: 6
Was liked: 65
Rank: now 1kyu-ish
Some background to Mizokami and also a pointer to an article by John F in "New in Go" that might help with this thread.

I think the concept of dividing the board into areas and counting the number of stones in each to determine "local numerical superiority" was introduced by Sonoda in his book "Good Points, Bad Points" published in 2004. Sonoda divided the board into two, horizontally, vertically or diagonally. If you have more stones in a given area then attack.

https://senseis.xmp.net/?GoodPointsAndBadPointsToPlay

Mizokami's book was published in 2013 May, he did not use the diagonals to divide the board, but I believe he took Sonoda's idea and turned it into a book.

John F posted "on a slew of books" in 2016 and referred to Mizokami's book (which is how I bought it) and "All about Sonoda's proverbs".

https://lifein19x19.com/viewtopic.php?f=57&t=13724&hilit=slew

Sonoda published "All about Sonoda's Proverbs" in 2016 May, in the second part of the book he analyses eight professional games and uses his proverbs to analyse moves at key moments, including "local numerical superiority" - but the areas he chooses are not as large as half the board - I am not sure how he chooses those areas.

There is an interesting article in GoGod's "New in Go" (Go Seigen_1.htm to Go Seigen_16.htm) where John F looks at two books and a seminar he ran with TMark back in 2002. One book is with O Rissei and the other is on games from 1998 to 2000 by Go SeiGen. Page 3 of this article shows eight areas of the board on the sides that are over looked in the mantra "corner, sides, centre" - but can have a tremendous influence on the game. On page 5 John F introduces the Go Seigen Group(GSG) - which is sitting in one of those blue areas. I think that in Sonoda's terms he would describe that group as "already alive stones" and Sonoda's proverb is "playing near already alive stones is small".

This article is in the light of this thread worth reading.

Back in 2016 John F said about Sonoda's book that "very, very happy to have found this book" - but that was before AI. Personally I like Sonoda's book and his advice on ways of looking at the board.

John Tilley


This post by John Tilley was liked by 3 people: Bill Spight, Knotwilg, SoDesuNe
Top
 Profile  
 
Offline
 Post subject: Re: X marks the spot
Post #22 Posted: Fri Jul 24, 2020 8:48 pm 
Honinbo

Posts: 9545
Liked others: 1600
Was liked: 1711
KGS: Kirby
Tygem: 커비라고해
John Fairbairn wrote:
Quote:
If there's no algorithm, then to me, I guess it's kind of interesting to read about, but not very useful. Things are useful to me if I understand them. Without an algorithm or a way to describe what's being talked about here in concrete terms, I don't have that understanding.


Let me recommend to you the other way of getting understanding: thinking. Cook for yourself instead of devouring fast food.


I *am* thinking. That's why I'm asking questions about the vague idea that you proposed. I asked about an algorithm, because you're the one proposing that you've made some sort of discovery. Just asking what it is. If there's nothing there that you'd like to discuss in concrete, understandable terms... that's cool, I guess.

_________________
be immersed

Top
 Profile  
 
Offline
 Post subject: Re: X marks the spot
Post #23 Posted: Sun Aug 02, 2020 4:20 pm 
Lives in sente
User avatar

Posts: 914
Liked others: 391
Was liked: 162
Rank: German 2 dan
Some incomplete thoughts:

“Corner, side, centre” has been proposed as outdated now. I think that it might be replaced by “gradually increasing corner areas”—up to 10×10….

Starting from there and looking at the other extreme: where is the first “corner area”, the one to perceive on the empty board? Certainly not 1×1. Most likely 3×3.

So, when you play 4-4, you're basically skipping the actual corner. (Yes, that seems “known” so far.) For what? Again, known: “for the outside”. That is: for larger parts of the further “rings” around the corner.

Classically, the sides (3rd line: the line of territory; 4th line: the line of “influence”), are valued highly. What I take from John's writings about the diagonals is that this is another line that should also be considered; maybe the “line of control”?

Of course, we wouldn't usually consider extending to the centre at first. But Go Seigen would, he proposed that three diagonal hoshi (hoshi, tengen, hoshi) might be the best sanrensei, and the big triangle (nirensei plus tengen) the second best. Alas, it doesn't seem to be that simple.

Finally some anecdotal data: when playing Katago online at 5 handicap, it would respond to a high answer to kakari almost invariably with this peep:

Click Here To Show Diagram Code
[go]$$B
$$ -------------------
$$ | . . . . . . . . .
$$ | . . . . . . . . .
$$ | . . . . . 2 . . .
$$ | . . . X . . . . .
$$ | . . . . 4 . . . .
$$ | . . . 3 . . . . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]

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

Top
 Profile  
 
Offline
 Post subject: Re: X marks the spot
Post #24 Posted: Tue Aug 04, 2020 2:08 am 
Oza

Posts: 3647
Liked others: 20
Was liked: 4626
Quote:
Classically, the sides (3rd line: the line of territory; 4th line: the line of “influence”), are valued highly. What I take from John's writings about the diagonals is that this is another line that should also be considered; maybe the “line of control”?


I have become increasingly aware that we may have become trapped into a Japanese Weltanschauung of go. I have had this feeling for decades, partly because of go-specific comments made by Go Seigen, partly through looking at old Chinese games, and partly from reading about Chinese strategy and tactics outside of go. But the feeling has recently intensified. This was driven in the first instance by trying to look at go though AI eyes (hence, in a way, this whole thread), but mostly by looking at old Chinese go again.

First, I think Chinese go has been misrepresented as "all-out fighting" or "power go". Even the Chinese have said this in the past, but I think too they are now re-evaluating this. But for us especially, who have different associations for words like "fighting". When we hear "all-out fighting" I suspect that the sort of images that swim up in most of our minds is pub brawls, riots, or bench clearances in baseball. Very, very few of us have had actual experience of "proper" fighting (I too was a post-war baby) and there our images are formed (or misformed) by Hollywood.

But if you go back to the classical era, "proper" fighting was something that was always a strongly potential part of life, and even scholarly Chinese gentlemen-officials who tended to despise the military men made sure they read and even memorised books like Sun Zi's Art of War. The Go Classic in Thirteen Chapters is even based on it.

It has always been difficult to get to grips with Japanese go vocabulary because until the 20th century there were essentially no texts. Books of games and problems, yes. Books with words, no. But it was very different on old China. We have lots of old commentaries, often quite long, and by go masters not amateur amanuenses. They say eyes are the window of the soul. Well, so are words.

It is very instructive to look at old Chinese go words. First of all they have been uncontaminated by Japanese vocabulary (unlike modern Chinese go terms). The palette is very, very different from Japan's. Not just in what is there but in what isn't there. No thickness, no moyos, no tewari, no miai - at least not expressed in the Japanese way. But some of the less obvious omissions (or near-omissions) may surprise you. "Attack" is actually quite a rare word. Instead they say an awful lot that a move is "severe". Is an AI bell tinkling already? Invasions, believe it or not, are rare. They like splitting moves - the whole development of Chinese fuseki, culminating in early/mid Qing times, can be seen as refinement of the 9-3 splitting move (in the area of a Go Seigen group incidentally :)) - and they like encroachments. (Of course, invasions risk creating a separate group and nobody likes paying tax.)

But the one word that stands out is shi 势. Roughly translatable as power, it subsumes thickness and much else. So much else, that the US military has spent a fortune on trying to understand shi, which is still the basis of Chinese military strategy and political strategy - and much else! Can you imagine the Pentagon spending money on studying thickness to understand Japanese politics. Old Chinese go is real hand talk - not a pale Japanese imitation.

And Sun Zi is always there in the background. The most eye-opening moment in go for me recently was discovering Huang Longshi's Five Grounds - based of course on Sun Zi's Nine Grounds.

Now, if you look at just these two elements - shi and grounds through the prism of old Chinese go terms, you get a feel for the game that is very different from our traditional Japanese-based feel and one that, I believe, more accurately mirrors what is going on in AI go.

You need to read at least one big book to even get a tenuous feel for shi (Mott & Kim is the definitive one in English) but if I may be allowed to try to distil it in to one English word, it would be CONTROL. With some caveats - it is not just control in the sense of having your hands round the other guy's throat or other important anatomical parts but is also control in the sense of sheer overawing presence (and so, to go off at a slight tangent, this is why I keep saying that thickness with weaknesses is not thickness and that thickness is not thickness unless it functions as thickness - but function here means control rather than attacking).

Now consider this: one common name for the four starting stones of old Chinese go is "shi stones". Control stones if you will. And these stones are placed diagonally, so that each player starts with a LINE OF CONTROL, i.e. X marks the spot.

The other element of my attempt to model AI go was to map the board into overlapping areas. It seemed to work, although in a crude way. But I tried this before I discovered Huang's Five Grounds, and since then I have refined my thinking to reflect that. Think of grounds as a locale, an area where both players may be present and engaged in various ways (as defined by 'control' - battle areas, if you like). These grounds shift as stones are added, which makes them way, way beyond my ability to program, but one thing I noticed, by reading the old commentaries and analysing the words, is that time is an absolutely crucial element. Not in the ultra-crude way of sente and gote but in the wider (and subsuming) sense of initiative, in which you can play a gote move to keep the initiative (the Japanese gote no sente, of course, but bigger than that). The number of comments in old Chinese texts that describe moves as well timed or the in right order struck me as far higher than anything in Japanese texts. I think we tend to think of timing simply in terms of technique or tewari. In reality, as the old Chinese commentaries show, timing is about keeping CONTROL.

I have also been struck (as I was decades ago but didn't understand it then) by how often the old Chinese commentaries have recourse to terms such as 细,which gets translated in tone-deaf ways such as "meticulous", and 落得, also rendered in tonally deaf ways such as "is advantageous". I believe the go sense of 细 is better understood as "tactically precise", and once you get in that mindset you realise that this too is about keeping CONTROL (and, mutatis mutandis, defining the shifting grounds). 落得 seems to have made an impression too on the old Japanese players. They rendered it as 打得, both versions meaning literally "can play" but the reference is to the (early) timing of forcing plays. Rather than the move being "advantageous" it is a no-loss move, which has very different connotations. And, in my new view of go, what that translates into is a way of defining GROUNDS. More bars tinkling on the AI Glockenspiel?

So, trying to put all that into a broken nutshell, I think a new theory of go in the AI era should look back at the old Chinese past (and/or Edo go, especially Dosaku, in Japan - but China mainly because of the huge volume of commentaries) and should emphasise primarily CONTROL and GROUNDS, with a strong secondary emphasis on severity, precision and timing as a way of keeping control and defining grounds.

And if you do revisit old Chinese games, think of them as battles and not fights. Even changing just one word changes the word associations and thus the perceptions. They do have highly developed strategy (as well as impeccable tactics!).

Top
 Profile  
 
Offline
 Post subject: Re: X marks the spot
Post #25 Posted: Tue Aug 04, 2020 8:36 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
Some time back I proposed that one of the oldest concepts in weiqi was that of the base, in particular that of the two space extension on the 3d line, the prime example being 8-11 (or equivalently, 9-12) base on the side. The concept may have been quite specific. Implicit in the 9-3 play is a notion of miai, in the sense of the ability to make a base on either side. There may also have been a notion of sente. But, as John points out, none of these concepts seem to have been articulated in words. Anyway, below are some ancient games from the GoGoD collection where I have made a few comments. :) BTW, it would be interesting to review some ancient games with AI, eh?








_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Top
 Profile  
 
Offline
 Post subject: Re: X marks the spot
Post #26 Posted: Wed Aug 05, 2020 3:50 am 
Lives in gote

Posts: 617
Liked others: 154
Was liked: 117
Rank: OGS ddk
KGS: Ferran
IGS: Ferran
OGS: Ferran
Maybe it's a silly question, but... if we started using the old 4-4 pre-placements... how many 3-3 early invasions would we get? Because every era has its favourite joseki, but I'm really getting tired of these ones.

Take care.

_________________
一碁一会

Top
 Profile  
 
Offline
 Post subject: Re: X marks the spot
Post #27 Posted: Wed Aug 05, 2020 8:43 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
Ferran wrote:
Maybe it's a silly question, but... if we started using the old 4-4 pre-placements... how many 3-3 early invasions would we get? Because every era has its favourite joseki, but I'm really getting tired of these ones.


Well, if you are Black you could play something like this.

Click Here To Show Diagram Code
[go]$$Bcm5 Only one invasion
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . 4 . . . . . . . 5 . . . . . |
$$ | . . . O . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 3 . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . 2 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . , . . . . . O . . . |
$$ | . . 6 . . . . . . . . . . 1 . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]


Here is a pre-AI game record that starts off with the traditional setup. The SGF file is a trimmed and edited version of one of the Elf GoGoD commentaries. I added estimates of the percentage loss of plays. Any play with a loss of less than 3% should be playable. A play with a loss of more than 5% is probably an error. IMHO. :) It should offer you some ideas for alternate plays and joseki.

Edit: Oh, yes. In the comments the decimal fraction is Elf's estimated winrate for Black. The second line shows the number of rollouts for that estimate.


_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.


This post by Bill Spight was liked by: Ferran
Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 27 posts ]  Go to page Previous  1, 2

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group