John Fairbairn's review on Yi Ch'ang-ho on Evaluating Positions (YC)

viewtopic.php?f=57&t=8625

is detailed enough to allow a preliminary comparison to my book Positional Judgement 1 - Territory (RJ) and, to some extent, the used methods. Unfortunately, Shikshin's book The Theory and Practice of Analysis is not sufficiently easily available in Germany yet, so, for now, I cannot also compare that one (hopefully in a couple of weeks or months).

There are further books with 'analysis' or 'judgement' in the title but little related contents, other books with analysis methods besides other topics and yet other books with trivial counting exercise problems. Such books I forgo at the moment. Cho Chikun's book teaches only small parts (summary: reduce to settle boundaries, consider also mutual reductions, count in pairs of 2) of the theory of other books, and so is also ignored here, except for the hint that Cho and RJ use the same basic judgement tool of peaceful reductions in implied sente (Cho) or explicit sente (RJ).

***

COMMON ASPECTS OF BOTH BOOKS (YC+RJ)

- The major procedure is the same: determine separately either player's territory by assuming the opponent's sente and reductions. Consider komi. (Double sente shapes: let every player reduce, when it is his turn to do so.)

- (Preferable precise) judgement is possible also in the opening and middle game. Methods for that are explained.

- Additional territory by the attacker made during reduction of the defender's region is considered (subtracted from the latter).

- The attacker's or defenders privileges are properly considered. (YC: not always shown. RJ: always shown.)

DIFFERENT ASPECTS

- YC: Korean / Japanese / Chinese text. RJ: English text.

- YC: real invasions are sometimes used (to determine the territory intersections). RJ: peaceful reductions are almost always used.

- YC: a territory parameter Alpha is used for: unknown (the value Alpha is not determined but only stated) or not easily calculated local additions; unknown or not easily calculated local subtractions (favouring the opponent; global influence / development potential. RJ: local values are always calculated precisely and, in case of optional additional territory, with Half Territory; local subtractions are included in the calculations (by directly subtracting the counted value instead of first annotating an Alpha parameter); global influence / development potential is not considered in Volume 1 (but will be studied in Volume 2, presumably by expressing a precise value for influence rather than by calling it Alpha). RJ: special use for a parameter (I call the parameter X) for moyo application.

- YC: moyos can easily be too difficult and need a guess or an Alpha. RJ: Quiescence and Half Territory always allow a precise territory count of a moyo, but it can sometimes suffice to use an Alpha (when certain strategic principles apply).

- YC: in a 50%-50% case of an initially almost neutral region (where either the attacker or the defender can get a small extra endgame), 50% of the attacker's potential, additional territory is subtracted. RJ: I think YC is wrong here. Instead, 50% means that attacker and defender must be treated equally. Therefore, the initially neutral region is worth 0 points. [In an asymmetric case with a tight boundary, determine the neutral region's endgame Count, but this is mostly beyond RJ.]

- YC: A 3-3 point in the opening is 4 points. RJ: A 3-3 point in the opening is 8 points, because, when the attacker approaches, the defender is exceptionally allowed to extend to the other direction in order to maintain life.

- Sometimes, local counts differ by 1/2 point or 1 point, apparently because somewhat different assumptions for the nature of reduction sequences are made.

- YC: The defender's influence used for reductions of the attacker's regions are calculated together with the defender's territory. RJ: Calculated when the attacker's region is evaluated. (Both methods are essentially equivalent.)

- YC allows gote or sente and different methods, depending on the "shape" of a region and surrounding stones. RJ: sente and method(s) are used consistently in the same manner for every region on the board. (Here, the methods really differ! However, consistency of either method should yield similar evaluation results. One must not mix the methods here, because this would lose consistency. YC's philosophy is treatment of early judgement like an extended form of endgame. RJ's (and Cho's) philosphy is that sente allows a successive global move sequence for all regions of a player. The more endgame-like a local region is, the more YC's and RJ's methods approach each other.)

- YC: influence is difficult to express in figures. RJ (but not in Vol. 1): it is reasonably easy and sometimes very easy. (If YC means territory figures, we agree. What I consider "easy" is influence figures.)

- YC: Territory near thickness is determined by visual guessing, it seems. RJ: The usual principles apply regardless of shapes, so that also territory near thickness is reasonably precisely given.

- YC: in unsettled local positions, sente is important. RJ: in global positions with locally unsettled regions, quiescence precedes judgement.

- YC and RJ consider strategy, but the topic is broad. Only specific aspects of strategy are considered. YC does so apparently without principles, RJ uses principles.

- YC: deire values. RJ: miai values. (I think that miai values are more powerful, because they allow comparison also of unequal local move numbers, such as in a ko fight.)

- YC: In a reduction sequence, finish off each region, then proceed to a next region. RJ: Use a global, meaningful order of reduction moves.

- YC: Imagine a sequence, until visual line drawing is possible easily. RJ: Complete reduction sequence. (In practice, both methods are essentially equivalent.)

- YC: extra chapters for (more) pro game examples. RJ: every chapter has related examples (with professional games) already in it.

ASPECTS OF ONLY YC

- Alternative (very) rough counting.
- A few proverb-like guidelines.
- Study of positions by their location.
- Study of relative strengths of groups.
- [I cannot know if the chapter The Value of Influence contains more than Fairbairn's review reveals.]
- [possibly more]

ASPECTS OF ONLY RJ

- Many explicit principles for how to construct reduction sequences and where to play its moves. (YC seems to make implicit assumptions yielding often similar or the same sequences, but the (too wide) absence(?) of explicit principles in YC leads to inconsistent application to different examples and, IMO, tiny value mistakes.)
- Definitions, where necessary.
- Also considered: types of groups and regions, remaining boundary defense moves, gote endgames, basic endgame kos, quiescence (and everything related), fighting positions, principles and half territory for moyos, principles for prisoners, middle game kos, updating counts, convenient counting, territory vs. area count.

***

Conclusion:

More than one method of positional judgement exists. There are common aspects, different but equivalent aspects and also really different aspects, which may be consistent in one method but, for consistency cannot be applied in the other method. YC makes pretty excessive use of a parameter, while RJ attempts to calculate precisely whatever can be determined. Both books have an overall impressive level of sophistication.

***

EDIT: corrections of significant typos.

There is a common expression in Korea, "플러스 알파 (plus alpha)" or sometimes "쁘라스 알파 (プラス alpha)". It is a Japanese derived made-up English with uncertain origin, which means "something more or extra".

_________________
Wait, please.

Does such a "plus alpha" phrase automatically imply the nature of the alpha being territory, influence or a combination?

(Just curious: in Korea or Japan, is alpha the common parameter for an unknown extra size? In Germany and German mathematics, usually it is called X, when the parameter is dynamic; and A when it must be a constant. Of course, there are exceptions and further possible symbols for parameters, such as X, Y, Z, U, V, W or A, B, C. Indices use i, j, k,.. Numbers prefer n, m. Angles prefer small greek letters.)

"Plus alpha" is used to indicate "anything extra value". So "four points plus alpha" in Go should mean "four points and more (in territory, thickness, or anything else valuable in Go)"

In Korean mathematics, x is more common than the Greek alphabets. But they are both commonly used, and usually depends on situation. This might be similar in Japan, where the most popular mathematics textbook in Korea "수학의정석 (mathematical joseki)" is originated from, by permissionless translastion.

_________________
Wait, please.

On Japanese Wikipedia or goo dictionary they say that the phrase came about from misreading the original 'plus X' in English. In Japan it is a very common expression in daily language. In Go the nature of the 'alpha' has to be determined from the context.

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

Reference:
viewtopic.php?p=143527#p143527



(It is a book I might decide to buy, to see all diagrams. Not because I expect much, but just to know what has been written.)

Lee's book [1], according to the reviewed part: I see the move sequences and the related conceptual ideas of reduction and sente. So there is little to be confused about: such are sequences to determine current territory. The more difficult part for me to guess is how exactly the alpha business works, and how the alphas are interpreted / applied by Lee. I do not understand this (well enough) from the review.

Other professionals: there is no such thing as a uniform positional judgement used by all professional players past and present. Conceptually, Cho Chikun was my predecessor for PJ theory and very different from Lee. Since I have worked out Cho's conceptual PJ sketches, I do not share your fear that I would be confusing what Cho was doing:) WRT to Takagawa / Ishida, I have only GoWorld 41 as a source for PJ information. I am not sure whether the problems only books were written by pro or amateur players, but I am sure that those books teach little, if any, theory. The rest I have seen was endgame rather than PJ. So there are not that many other professionals for which I might be confused about their theory. Rather I wonder where are any further PJ books? You sound as if there were countless such books... Tell us.

Much of what I've seen of positional judgement is taught on video for minimum territory evaluation of an area to do estimations. What you were showing in another topic was adding a stone and changing the score which would make no sense to me.

Other books I've read are Yuki Satoshi's Positional Judgement book from Mycom and O Meien's book.

Until you specify exactly the diagram, I cannot guess which problem you are having with it.



Would you like to give more information on them?

viewtopic.php?f=15&t=8764



I may in the future, but I have limited time. I'm pretty sure O Meien's has been reviewed by John Fairbairn in the past. I picked up Yuki Satoshi's at a used book store fairly cheap, and thought it was good at explaining the concepts and giving practical problems to back them up. Mizokami Tomochika just released two months ago a new book for positional judgement in 30 seconds. If I didn't have enough in the queue I might check that out later, but Lee Changho's looked interesting after John's review of it.

I cannot guess which of the diagrams in that message you mean. None seems applicable for "adding a stone and changing the score".

Actually I think most (if not all) diagrams in that message satisfy this.

_________________
Geek of all trades, master of none: the motto for my blog mostlymaths.net