I'm trying to fill out parts of my collection while I'm in japan and I was wondering if anyone had any good books on endgame theory/tsumego they could recommend. I've been doing some googling and came across O Meien's Absolute Counting which seems to cover theory very thoroughly, but I'd love to hear some opinions on other books to consider.

I doubt if there is anything better than O Meien on theory, or even comparable. Otherwise I would go for endgame tesuji books.

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

Visualize whirled peas.

Everything with love. Stay safe.

Nice to know! Are there any tesuji books that stand out to you or should I just grab anything I see?

I'm not really up on the latest in go books. I did a quick search on amazon.co.jp and noticed a number of books that combine yose and life and death, which is a good pairing, IMO. I even saw 高川囲碁読本 第5巻 ヨセと死活 , which was the first book on yose that I studied, and I still like. It is out of print, OC, but apparently you can get it on Amazon. (I wouldn't pay 12,000 yen, though. It's not that good. ) I recommend the whole set.

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

Visualize whirled peas.

Everything with love. Stay safe.

There's actually a couple copies going on amazon for about ￥900 + ￥250 shipping. I also found ヨセの手筋―終盤の逆転力をつける by Cho Chikun and 有段を目指すヨセの手筋 for cheap, both having some good reviews online.

Is this the endgame book of O Meien that Bill recommends? Is the PDF without digital rights management so that opening and viewing it is easily possible without having to know Japanese? Is the PDF without javascript so that viewing does not create a security risk? Does it contain enough go diagrams to make reading meaningful without knowing any Japanese other than numbers, points, Black, White? (I guess I might figure out any Japanese annotation of fractions.)

The O Meien book is the same one. The current pdf/kindle books are a combination of the older book with some more information.

There are many yose tesuji books which are probably more generally helpful. A lot of books have started to become available for kindle which is nice.

The title is different, but the three pages you show are the same as in the book I have, even the page numbers.

As it happens, O Meien makes a small inaccuracy on page 46. He does not have the concept of ambiguous plays, and treats as gote instead. gains 3 points, but so does the Black reply, Ba - Wb, Bc in diagram 47. That means that the count for Black's corner should be 12 points, not 11.5.

Surely this would not lead O to make a mistake in a game, because he would give White on the left a half point less territory, as well, yielding the same net count.

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

Visualize whirled peas.

Everything with love. Stay safe.

There's a couple of things I don't get about this, Bill, if you wouldn't mind elaborating.

First you O doesn't have the concept of ambiguous plays. I'm not really sure what that means here, but in the problem just before O specifically says that the usual sagari-sagari trick doesn't work here. To be precise he first says : "This sagari-sagari is fundamental in counting territory, but is limited strictly to cases where the conditions for both sides are the same.
Next we deal with cases where sagari-sagari cannot be used."

Then he adds: "Unlike the previous problem (Q) the hane of White A is no longer sente. I suppose you have already noticed that we cannot use the method of sagari-sagari with Black A, White B (but were there a Black stone at C we could do so without difficulty). However, even in this case it is possible to produce a definite number of the type 'star-symbol' points. Because it produces precise figures for cases that hitherto have been treated vaguely, I would like you to really master this method."

What that specifically refers to is not important. I quote just to show that he fully acknowledges the issue (I think) - he just doesn't give it a name. Am I right?

Second, since this (Problem S) is a count of Black's territory, I don't see how you can add the extra half point for Black's possible future crawl. That would be valid only if you trying to give the move a value (which is the topic of his next chapter), surely? If we just treat the visible portion of White's putative territory on the lower edge, it will be either 4 points or 3 points, i.e. a territory count of 3.5 points for him, so the relevant intersection is fully accounted for by O's method and Black should indeed count his own area as 11.5. If you say 12 for Black you have to say 3 for White, no?

However, in the previous Problem R, O does add a half point to the Back territory in the way you have done, so it's clear that there is indeed a "minor inaccuracy" somewhere. It's just that I would have believed I had a clearer grasp of things if you'd said the inaccuracy was in Problem R! If the explanation is that this involves "privilege", then we could add sloppy writing to the criticism as he only defines this later, although this sort of deferral is normal Japanese style, as exemplified also in the Japanese rules. However, we would not then be talking about reducing White's count by a half point, but by a whole point, surely.

OK, John, here goes.

Click Here To Show Diagram Code

[go]$$W Diagram 47
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . O X X X X O c . X . |
$$ . . . b 1 a . . . . . . . |
$$ --------------------------[/go]

O says that after , Ba - Wb, Bc is gote (Me: True, or reverse sente), so Black does not reply (Me: Does not follow).

Click Here To Show Diagram Code

[go]$$W Diagram 48
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . O X X X X O 2 . X . |
$$ . . . . O 1 3 . 4 . . . . |
$$ --------------------------[/go]

O says that we may consider - to be White's privilege (Me: Not in this case). And that Black gets 9 points of territory (Me: True).

Click Here To Show Diagram Code

[go]$$B Diagram 49
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . O X X X X O . . X . |
$$ . . . a 1 . . . . . . . . |
$$ --------------------------[/go]

O says that after Black gets 5 more points than in the previous diagram, for 14 points (Me: Correct). He then says that these two possibilities are 50-50, and then calculates the count as the average of 9 and 14, or 11.5 (Me: Slight inaccuracy). On this page O says nothing about "a", but it is plainly Black's sente, so we consider that Black gets to play there with sente. That is why O's inaccuracy will not affect the difference in White and Black counts.

Plainly O is treating the Black sagari and the White hane as gote. That is why I say it is a slight inaccuracy. The White hane is ambiguous between sente and gote. As I say at http://senseis.xmp.net/?AmbiguousPosition , in the go literature you sometimes see ambiguous positions referred to as sente and sometimes as gote. In this case O says that it is gote, and makes his calculations accordingly.

Despite treating as gote, O's calculations will come out OK. He will calculate that it reduces Black's territory by 2.5, on average, and increases White's territory by 0.5, on average, for an average gain of 3 points.

O says that after Black will tenuki and that White will have the privilege of playing the sente sequence in Diagram 48. Why are those statements inaccurate?

Click Here To Show Diagram Code

[go]$$W Diagram 47a
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . O X X X X O 4 . X . |
$$ . . . 3 1 2 . . . . . . . |
$$ --------------------------[/go]

They are inaccurate because Black can play - without loss. On average, OC. What is Black's territory after ? 12 points, which is 3 points more than in Diagram 48. That means that - gains exactly the same number of points as . And that means that the result is exactly the same as in the original position. So we may regard Black's original count as 12 points.

White gets the privilege of continuing with sente only if Black does not reply to . But why shouldn't she reply, since she gains as much as White did with ? Oh, there may be circumstances where Black will play elsewhere so that White can continue locally with sente, but for Black to reply locally is normal.

Like the sagari-sagari trick, for Black to reply to preserves the original count, but unlike the trick, it is normal play.

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

Visualize whirled peas.

Everything with love. Stay safe.

Thanks everybody for help! I have decided to order softcover from Amazon Japan. Enter the ISBN there, be sure to distinguish softcover from Kindle and Amazon service from third sellers, use the website like every other Amazon website, ensure to have set some payment method with current data, if necessary use Google Translate or another translation service, be careful not to subscribe to Amazon Prime accidentally. Price of the book ca. €12.50, airmail shipping to Europe (ca. 1 week) ca. €9.50, on arrival some customs fee might apply. The shipping fee contains some hefty Amazon margin but the price of the book is ok, so the total is reasonable.

I do not expect to learn anything new from the book, but I want to know what the allegedly best Eastern endgame book is. Maybe the author surprises me and has to offer some surprising contents:) Looking at the samples, the diagrams show too many empty intersections wasting space but at least numerical calculations follow international standard so it should be easy to understand at least the numbers and their meanings as either move values or counts, presuming a previous understanding of the basics of modern endgame theory / absolute counting.

Thanks, Bill. When I don't feel comfortable in understanding something, I tend to take refuge in literalness (which does mean trusting the author, of course). Since O stresses "absolute" I automatically reject the fuzziness of "normal" and "circumstances." Despite your kind explanation, my comfort zone hasn't really expanded - I still regard Problems R and S as being treated differently by O, and while I can sense why, I don't see that clearly enough to apply the principle to other positions on my own. Which is "normal, " of course!



Robert: "ambiguous" here too. Do you mean that you think you know it all already or that you won't be able to follow enough of the text? I'd be very surprised if you believe the former, at least in practical terms. Although much of the discussion here has been more or less about theory, I don't think his book is an attempt to define a new theory. I see it as an attempt to strip away the layers of grime and varnish that have obscured the original Japanese method. That's precisely what his introductory chapter is ALL about - showing how a couple of amateurs screw up a simple count by applying the wrong words/concepts. But in a way that's a sideshow. For me the real eye-opening part of the book (and the bulk of it) was the way he applies the restored method to a decent-sized series of whole-board early endgames where practical decisions have to be made. He shows exactly how and why he makes these decisions, and in particular his Calling the Election formula is a novel contribution which I don't really believe you are likely to have reproduced. Although it's empirical it works down to the level of half-point endgames and so it's very close to the theoretical level. It would just fail (or be too difficult to apply) in some rare ko/seki cases. It also has the merit of being very fast to apply.

I cannot know yet what the Calling the Election formula might be. Since you say that it is for the early endgame, it might involve factors I have not been studying yet, such as translating influence to components of endgame move values. So if it involves approximation fuzzier than expected averages of purely territorial values (without or with some error margin), I have not studied it yet.

However, I have studied late and early endgame (relying on per move values) for more than half a year now including proofs for many (70+) propositions / theorems so will surprise you with new discoveries. OTOH, I am working carefully and spend, say, a dozen propositions on a topic to prove every aspect instead of building dreams on air. So I do not cover too many topics. However, what I study is shape-independent and value-tolerant.

The more I study the endgame the richer it unfolds. So much is hidden and only discovered on very serious study. Therefore, if O Meien has not just written down basics of modern endgame theory but also discusses some topics requiring his or others' very serious theoretical (formal or not) study, it is likely that his and my theory do not only cover the same topics. If so, the samples of his book would be misleading.

I've just read up to this point (decided to get the kindle version) and I'm not sure why you think they're treated differently. In both cases (problems R and S) he assumes neither player will respond locally to the initial boundary play with a move that ends in gote. This is why white can play the crawl in sente and black has a sente push at a. I assume the reason he doesn't show the move at a played on a diagram because this is very easily calculated using the techniques from the first section of the chapter.

Personally I'm not sure why Bill feels there's an inconsistency because the move at a seems like a feature not a bug of this counting method. But on the other hand this is my first venture into learning how to count boundary plays so either I have the benefit of being unencumbered by other random bits of knowledge or I'm too naive to really appreciate the situation. Either is very possible, I don't want to claim to understand more than anyone else in this thread given everyone's experience.

Yes, that's how I originally took it. But Bill is saying that similarity of treatment masks a slight inaccuracy by O. Which means they should be regarded as different

I guess I just don't really understand why this is an inconsitency. Unless there are situations in which the count by O's method comes out differently then the count is just as useful a tool for deciding where to play. Of course if it's just a local problem then black would immediately respond but I was under the impression counting is used to make comparisons between several moves on the board.

This is true with every topic, isn's it ?

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

For every answer, I get several new questions, o o...

Here is the difference between the two.

Click Here To Show Diagram Code

[go]$$W S
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . O X X X X O 4 . X . |
$$ . . . 3 1 2 . . . . . . . |
$$ --------------------------[/go]

gains 3 points, and - gains 3 points. Net result: 0 gain.

Click Here To Show Diagram Code

[go]$$W R (Black follower)
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . . O X X X 4 . . X . |
$$ . . . . 3 1 2 . . . . . . |
$$ --------------------------[/go]

gains 2 points, but - gains only 1 point. Net result: 1 point gain for White. Black territory: 10 points.

Click Here To Show Diagram Code

[go]$$W R (White follower)
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . . O X X X 4 . . X . |
$$ . . . . 7 1 3 5 6 . . . . |
$$ --------------------------[/go]

gains 2 points. Then - gains 1 point. Net result: White gains 3 points. Black territory: 8 points. Average Black territory after : 9 points.

O Meien shows none of these diagrams. For R he shows this diagram, for the result after .

Click Here To Show Diagram Code

[go]$$W Diagram 45
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . . O X X X 2 . . X . |
$$ . . . . 5 W 1 4 . . . . . |
$$ --------------------------[/go]

This is not a real life diagram. If White plays , except as a ko threat, he will continue to push in. This is like the sagari-sagari trick. Both sente exchanges, - and - , gain nothing on average. Result: 0 gain from the position after was played.

Edit: O Meien uses the sente-sente trick for position Q.

Click Here To Show Diagram Code

[go]$$W Diagram 42
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . . . O X X 2 . . X . |
$$ . . . . . . 1 . . . . . . |
$$ --------------------------[/go]

This diagram shows White's sente, which gains nothing on average.

Click Here To Show Diagram Code

[go]$$B Diagram 43
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . . . O X X X . . X . |
$$ . . . . . 2 O 1 . . . . . |
$$ --------------------------[/go]

This diagram shows Black's sente, which also gains nothing on average. Net result: 0. So we count Black's original territory as 9 points.

O uses the sente-sente trick for the original position in Q and for the position after in R. But there is no sente-sente trick for S, neither in the original position nor in the position after . O does treat S differently from the other two.

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

Visualize whirled peas.

Everything with love. Stay safe.

Hmmmm. I overlooked this before, but actually position Q is ambiguous, as well, but in a different way.

Click Here To Show Diagram Code

[go]$$W White sente
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . . . O X X 2 . . X . |
$$ . . . . . . 1 . . . . . . |
$$ --------------------------[/go]

White can play as sente and then play elsewhere.

Click Here To Show Diagram Code

[go]$$W Gote
$$ . . . . . . . . . . . . . |
$$ . . . , . . . . . , . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . . . . . |
$$ . . . . . . . . . O . . . |
$$ . . . . . . . . . . O O O |
$$ . . . , . . . . O O X X O |
$$ O O O O O O O O X X X . X |
$$ . . . . . O X X 2 . . X . |
$$ . . . . . 5 1 3 4 . . . . |
$$ --------------------------[/go]

Or White can play this sequence, ending in gote.

The ambiguity is clearer, perhaps, with the method of multiples. The SGF file shows play in four copies of Q. Note that when White plays first he should play every one as sente except at the last, when he should take gote. When Black plays first it doesn't matter whether White plays gote or sente. It's ambiguous.

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

Visualize whirled peas.

Everything with love. Stay safe.