YI CHANG-HO ON EVALUATING POSITIONS

John Fairbairn and GoGoD, 2013. For use only on the Lifein19x19 forum and the GoGoD CD.

INTRODUCTION

The following is based on a book by Yi Chang-ho [Lee Changho]. It should be considered as a

review but I have done this in a fairly extensive way so that some of the information here may prove
useful even to those without the book. It is not a tutorial, and my aim has been mainly to hint at what
is available in the Oriental go world but is not yet available here. Nor is it a translation, or even close
to one. It may passably be called a summary, but if so it is a brutal summary, and in particular I have
shifted the emphasis at many points. I have omitted certain things altogether, and have also added
some remarks. However, I hope I have been faithful enough so that those who do acquire the book but
cannot read the original languages (Korean or Japanese) will here have enough guidance to the text to
make good use of the diagrams in the original.

The version of the book I used is in Japanese: by Yi Chang-ho 9-dan (Seibundosha

1999, ISBN 4-416-79948-9) but the original was in Korean (, published by Samho Media
Co.). I have no idea about current (2013) availability but I strongly recommend getting hold of a copy.

How much input Yi personally had into the book I cannot say. The usual arrangement is for a ghost
writer to interview a professional one or more times and to ask for a check of the final draft. It is up to
the professional how scrupulous he is. The nature of this book, with many specific details, indicates at
least a fairly significant input by the pro.

A NOTE ON TERMS

The phrase (and its Korean equivalent) means literally positional assessment but since
that is nearly always accompanied by counting, it is often translated as counting. Here, however, I
will stick with evaluation. Counting will be reserved for the term and to count (territory) for
.

The terms (framework) and (territorial framework) are often treated as synonyms when
rendered into English (often as moyo) but the latter lies between a framework and actual territory.
Here I have used the term framework for and prospective territory for . The latter is
the important term in this context. Other translation points will be discussed in situ.

FORMAT OF THE BOOK

The book has about 250 pages (with very many diagrams) and four main sections.

Section 1 FUNDAMENTALS OF EVALUATION is about 60 pages long and covers eight topics. What is
given here is mainly a digest of only some of these topics, which are:

1. How to count territories

2. Prospective territories in the corners
3. Prospective territories on the sides
4. On development potential
5. The essentials of de-iri counting
6. The value of influence
7. The relative strength of groups and how this changes the sizes of territories
8. Real games! The essentials of evaluation in practice

Section 2 TRAINING FROM ACTUAL GAMES is also about 60 pages long and is a set of eight
positions to show how the fundamentals of Section 1 are actually applied to real positions. One
example is summarised here, and the titles of the others are given.

Section 3 PRO EVALUATIONS is similar to Section 2, but pitched at a higher level and the six
examples are treated in more depth by being divided into three themes. This section is ignored here,
but the headings will be given. This is mainly for those who need help with the original book, but it
will also hint at the other aspects of evaluation that need to be taken into account as you progress in
your studies.

Section 4 EVALUATION TESTS is not treated here. Ten problems are given with a scored choice of
moves.

There are some also quizzes on boundary-play counting in standard local endgame positions
interspersed throughout the book. These are ignored here. There is a discussion of five ways to
improve your evaluations. The gist of that of that will be given.
SECTION 1

(1) How to count territory

The average amateur counts locally, but an overall evaluation is more important. And more important
than counting pure territory is its latent value (development potential, power, thickness).

Yi says: Counting in the endgame when the boundary lines of territories are virtually fixed is not
difficult. Instead the focus of his book is on counting incomplete prospective territories in the
opening and middle game.

The fundamental method is to draw a line between stones that are up to two spaces apart (jumps or
knight moves) or between the end stones of a group and the first line (edge). This can be justified by
imagining forcing moves made by the opponent. Sometimes the opponent will also have follow-up
forcing moves and so the outline of the prospective territory may have dents in it.

This, however, produces only an approximate value valid only under laboratory conditions. This
value has to be modified in the wild when the opponents predator stones are close by.

Diagram 1 is a simple example. Note that diagram numbers here bear no relation to the original (1
here is 9 in the original) and I have even changed many diagrams for succinctness.

Diagram 1

The Black area here is counted as a prospective territory of 8 points. The boundaries are assumed to
be the points marked X, these being the points at which Black would play if White made the most
facile forcing moves. If there had been no White stones to the left, the area would have been counted
as 10 points, with Xs at B and C, and also at the circled point. But because White is both present and
strong, we can assume that he will not only force at A but can also play at C in sente, and so the
guesstimate is reduced accordingly.
 Diagram 2

Diagram 2 is a more complex example where Whites strength at the right also has to be taken into
account, since he can now play moves such as White 1. The details of how to handle this sort of
position are covered in Section 7 The relative strength of groups and how this changes the sizes of
territories. But a very large proportion of positions can be handled by the method shown in Diagram
1.

(2) Prospective territories in the corners

Diagram 3

Can you count a prospective territory for one stone? Yes. Komoku, with Xs down to the edges as in
Diagram 3, is 6 points approximately. Adding a small knights move to it at A and shifting the
boundary on the lower side to the circled points makes it 11 points approximately, but with the
difference that the shape now acquires development potential (my rendering of ), i.e. the
possibility of future growth, mainly up the left side here, and so this sort of position is better described
as 11 points + . Section 4 On development potential deals with this.

On this basis, it might seem that a large knights move added to komoku is worth even more 13
points. No! First, it is thinner (= flimsier/weaker) and White can invade, so the territory is not at all
secure. Second, the nature of the large knights move is to shift the development potential more to the
lower side, where growth towards the centre would lead to a shallow tray rather than a deep box, and
so is worth less.
In the case of a high corner enclosure (at the point above A), the prospective corner territory is even
less secure, but as the development potential up the left side is even stronger, the becomes
correspondingly more valuable.

Diagram 4

Further, if we assume the plausible sequence of Diagram 4, Blacks prospective territory may seem
like 13 points, but in the process he will almost certainly have to concede White the forcing moves of
1, 3 and 5. We do not assume that White then immediately plays 7 but we can assume he has a 50-50
chance of playing it, and so has a 50% stake in the 4-point territory outlined by the circles. We can
count this as 2 points and so Blacks area is really worth 13 2 = 11.

Diagram 5

In Diagram 5, Blacks prospective territory is counted as 7 points. This is not reduced by Whites
territory because no one-way forcing moves are involved (i.e. Black may get to force first), and so we
compare simply on the basis of two separate groups, i.e. count each independently.

As regards the other standard corner moves, a stone at the 3-3 point is counted as 4 points and is
assessed as also having high security.

A stone at the 4-4 point is not 9 points, because White can live at the 3-3 point. It is best to count it as
6 points, or 8 points if it has good development potential.
 Diagram 6

It is important to learn the typical or standard moves. Blacks area is counted as 13 points, not 14,
because when White plays 1 Black has to answer at 2 not A.

Diagram 7

Being played against on both sides is a cardinal sin in joseki, and if it happens the count of
prospective territory has to be reduced accordingly. In Diagram 7, Blacks position before White 1 (i.e.
with three starting stones) is not counted as 16 points. Because Black has allowed himself to be
bullied, this is counted as 3 points for him. This is because he faces suffering both White 1 and A as
forcing sequences, in which case his corner territory is 7 points, but as White gains the 4 circled
points in the process, we have to count 7 4 = 3. If it is Black to play and he adds a stone at 4, we still
count it as approximately 14 2 because White can gain a couple of points by forcing below 7. All
the above is, for simplicity, without mentioning the left side, where White can gain even more.

(3) Prospective territories on the sides

 Diagram 8

Black 1 in Diagram 8 in isolation is bad shape. The reason is that if White plays A, Black has to
defend at B or C to prevent White D, and then Black 1 ends up misplaced. Black 1 is only a good
move if it works as part of development potential, for example if Black has stones at the triangled
points.

Diagram 9

Blacks two outside stones in Diagram 9 are usually counted as 3 points because of the standard
exchange shown. But they are sometimes counted as 0 points. The point is that if he gets this position
for whatever reason, he has to give higher weight to adding extra moves for security. And the main
message of that is that extending too far does not mean more territory.

Diagram 10
In contrast, in Diagram 10, Black can count his prospective territory as 17 points, but his position is
actually overconcentrated and inefficient. For efficiency, his triangled extension should have been at
A. But how would we then count that position. Its hard, but since we know from experience its
better than the overconcentrated version we can say it is worth 17 points + . Here stands for the
richness or fertility endowed by the addition of development potential. The message in brief is: add
extra for development potential.

By analogy we can also count big-point extensions from other corner enclosures as 17 + .

(4) On development potential

Development potential means added value. The important lesson to learn here is how development
potential eventually turns into territory.

Diagram 11

Blacks position in Diagram 11 before White 1 has little likelihood of being reduced significantly by
White we can assume the moves shown are Whites best effort and so we can count Blacks
prospective territory as 15 points. But it also has added value where?

Diagram 12

Black of course has some development potential up the left side, but the main development position
here is Whites, to the north and east of his position. Since Black can reduce or eliminate that as
shown in Diagram 12 without any impact on his own position to the left, this can be counted as added
value of the position on the left.

Diagram 13

If Black has strength on the right he can even invade. After Black 1 in Diagram 13, the invasion at A
is threatened, which may force White to jump to 2. However, it is possible not just to take away
development potential from the opponent but to give it to him, and Black 1 is double-edged even as a
forcing move because it gives White considerable added value in the centre. Actually, Black 1 may
not even be forcing. There are standard sequences if White skips 2 and lets Black invade at A where
White ends up with no territory but more influence facing the centre, which may be what he intended
all along.

Diagram 14

What about the common position in Diagram 14? Black may look worse off 5 points against 7
points for an equal investment of three stones but he may even stand better (the value of A here is
not the issue that is discussed in Section 5 The essentials of de-iri counting). The point is that
Black has an element.
 Diagram 15

Black can force at 1 and 3 and so acquire development potential (northwards) in a way that White
cannot.

Without going into why here, sanrensei and similar three-stone positions on the side (e.g. the Chinese
fuseki) are counted as 25 points with some similar thinking.

Diagram 16

Incidentally, if Black plays 1 in Diagram 16, his position is counted as a minimum of 30 points. If
White pre-empts this and wedges in at A, the move White A is worth about 20 points.
 Diagram 17

Going back to the position at the beginning of this section, in isolation Black 1 of Diagram 17 is
overconcentrated and inefficient, and Black A instead would promise the same prospective territory
but have the of development potential. However, if White is strong round about, Black 1 is almost
certainly the best move. Not every position has an .

(5) The essentials of de-iri counting

This section is essentially omitted here, but one case is worthy of note as a reminder that we are here
talking about evaluation in the opening and middle game, not the endgame.

Diagram 18

You may be tempted to count the Black position in Diagram 18 by visualising ordinary de-iri (profit-
and-loss) moves on the side, such as Black 1 etc contrasted with White playing first on the edge. You
may then falsely count Blacks prospective territory as 17 points. There are two problems with this.
One, Black 1 is gote. Two, this is not the endgame. So long as it remains the middle game Black has
to allow for the possibility of a fairly standard invasion by White. Wherever such a move (a or
move left behind) remains, as in Diagram 19 below, you have to allow for it, and to allow for the
fact that playing here is gote, Black counts the average of the two cases.

Diagram 19

This is not gote for White, of course, but it may be premature and so may not be sente either, so we
assume gote-gote as in standard de-iri practice. This is then counted as 11 points for Black, giving an
average prospective territory of 14 points. In passing, the initial triangled move, so popular with
amateurs, can be seen not to be as good as it feels.

(6) The value of influence

By influence here we mean (power, outreach - not thickness, solidity). Rather than thickness,
this is the true opposite of or profit. It refers to shapes that can extend their strength and
development potential in the centre. Secure profit is like cash. Development potential is like a
promissory note which can bounce it has a hidden risk. It is therefore difficult to put this latent
potential into figures.

Diagram 20

In Diagram 20 Blacks prospective territory is counted as 5 points + , but here means strong
influence. As a simple example of what that can mean, if White develops a strong position around A,
Black will be hard pushed to get more than his basic 5 points. But if Black can add a stone at A he can
then count his prospective territory as approximately 17 points, and look forward to even more
development potential.

An important point to note is that when Black adds to his position, the nature of influence is such that
he can add not just prospective territory but even more development potential. By taking risks he can
become rich rather than just well off.

(7) The relative strength of groups and how this changes the sizes of territories

Strengths and weaknesses are very important. Having no weak points and solid shape means being
thick. Having flimsy shape means being thin. Being thick is especially important when games are
close. With similar shapes or potential the thicker player will win.

Diagram 21

On the face of it, in Diagram 21 Black has 19 points on the right and White has 4 points. But Black
has defects at A and B, while Whites base can be undermined so that he is made to run away. Counts
only apply when things are stable. In such mutually unstable cases, sente is important.

Diagram 22

If Black plays as in Diagram 22 he can actually count 25 points.

 Diagram 23

However, sente does not solve Whites problems. He made a mistake earlier on and allowed himself
to be played against on both sides. He still has to pay for that mistake. Playing White 1 eases the
pressure, but after Black 4, Black has a prospective territory of about 22 points and an aim at A, and
so Whites first move has gained little. Earlier on, White should have given priority to pre-empting
Blacks triangled extension by extending into that area himself.

To see how much difference this makes and why the sin of being played against on two sides is so
heinous, consider Diagram 24 where White has correctly upstaged Black.

Diagram 24

Since Black A would be gote, White has every expectation of being able to play White 1 as well as
the sente boundary plays on the left. White has therefore added 11 points to his previous total of 0
points. Blacks area has been reduced to 7 points. The de-iri value of Whites triangled move is
therefore over 20 points. This is an example of the less talked about middle-game de-iri rather than
endgame de-iri.
 Diagram 25

Diagram 25 illustrates another concept covered in this section of the original. White cannot count
anything yet for the area on the side because it is weak on both sides and had no external help. For
example, White 1 is countered by Black 2. That is, White may get some territory to the right but
Black will make countervailing gains on the left, and vice versa.

(8) Real games! The essentials of evaluation in practice

Three methods of evaluation are used in practice, depending on how much time is available.

(a) a precise count this is used mainly in the middle game and endgame;

(b) rough counting, say in units of 10 mainly used in the opening;

(c) comparisons of similar positions and offsetting one against another mainly for the opening and
middle game.

All three methods are illustrated in the following example, which is, incidentally, assessed in terms of
Whites privilege then Blacks privilege rather than Black to play and White to play, a subtle
but useful difference.
 Diagram 26

(a) Precise method

The procedure is:

(i) Make Whites privilege plays so as to adumbrate Blacks prospective territories.

(ii) Make Blacks privilege plays so as to adumbrate Whites prospective territories.
(iii) In cases of double sente, count them for each player.
(iv) Add in komi.

To wit:
 Stage (i) Whites privilege

The upper side is not assumed to be double sente because a Black play in that area threatens both to
rescue his triangled stone and to kill the White group much bigger threats than White can muster, so
we can assume Black will get to play here first. (The left side is, however, a double sente case.)

On this basis we can count the Black prospective territory on the left as 25 points and the one on the
centre right at 37 points, for a total of 62 points.
 Stage (ii) Blacks privilege

On this basis, we can count 10 for White on the top side, 4 in the top right, 12 in the lower right and
42 in the lower left a total of 68 points.

It is Black to move. Even without komi, White is ahead by over 5 points on the board. Black therefore
has to do something desperate. The evaluation tells him that his next move must be invasion of the
lower left and not a namby-pamby reduction.

(b) Rough method

The rough method is similar to the precise method but as it used when time is short it is based on a
quick-look guesstimate instead of laboriously trying to visualise both sides privilege plays. By all
means uses precise counts for small, easily assessed areas, otherwise guess. Obviously experience will
improve the guesses, but you may be surprised at the guesstimates given in the book for the above
example. Remember they are based simply on a quick glance rather than a sloppy way of doing a
privilege count. Blacks areas are counted as 25 and 30 (instead of 25 and 37). Whites areas are
counted as 10, 4, 12 and 30 (instead of 10, 4 , 12 and 42), the 10 here being a quick-glance guess
rather than based on privilege plays calculation. A weaker player could also count the lower right as
just 10 but corners tend to be standard shapes that good players can memorise.

So, White is still ahead, but hell be even more grateful for the komi in this case.
(c) Comparison method

Using the comparison method, which is even faster and cruder because it attempts to omit some
arithmetic, Blacks top centre area is offset by Whites lower left, while Whites top area and his
lower right together cancel out Blacks top left area. That leaves White ahead by the size of the top
right plus komi.

SECTION 2

Eight positions are discussed to illustrate the principles of Section 1 as applied to real positions. The
themes are:

1. Evaluations in the opening and middle game

2. Whether to make territory or to attack
3. The way to make use of influence
4. To defend pre-emptively or to counterattack
5. Do-or-die moves and how to handle them
6. Forking points when there is a selection of moves
7. To invade or erase?
8. Evaluations in the endgame

As you will see, several of the themes are dichotomies. Decision making is called for. That is why
evaluation is so important.

SECTION 3

This is similar to Section 2, but pitched at a higher level. The themes are as follows, based on three
different positions opening, middle game and endgame - from each of six games. The positions are
discussed to a large extent on the basis of the x-marks-the boundaries method.

1 (a) Erasure and counterattack

(b) Thickness and managing the centre
(c) Dealing with critical situations
2 (a) Whole-board rather than local positions
(b) Wrong direction of play leading to a reversal
(c) Do-or-die moves and half-point wins
3 (a) Large-scale vision is important
(b) Attack and defence revolving round boundary lines
(c) Upsets in the endgame
4 (a) The vision to discover large-scale urgent points
(b) The ability to seize sente in a flash
(c) Overplays that miss the easy path to victory
5 (a) The timing of invasions and their preparation
(b) Single mistakes in the order of moves
(c) The relationship between evaluation and reading
6 (a) Different widths of extensions according to the position
 (b) Large-scale vital points that defend thickness
(c) A double-upset loss that invited optimism

It says in this section that when a pro is thinking for a long time, he is not being taxed by the difficulty
of reading. It is because he is weighing up the evaluations of the different positions.

SECTION 4

Omitted here.

FIVE BEST WAYS TO IMPROVE YOUR EVALUATIONS

This is a tailpiece. The gist is:

1. Count territory in detail at least thrice in a game (transition from opening to middle game, transition
from middle game to endgame, and any critical points where the outcome is uncertain).

2. Do an evaluation whenever you have to make a decision

3. Think of all territory as prospective territory only.

4. Check at all times to see there are no weak points.

5. Learn by heart the standard prospective territory shapes and the sizes of their boundary plays.