Exchanges (part 2)

A Double-edged Sword
by Charles Matthews
This article first appeared in the
American Go Journal
For a background shape topic, Im looking this time at angle plays such as White 1 in Diagram 1. Black will answer
with an extension at 2 on one or other side. Then White should play once more in this part of the board, or there is
a chance that the exchange of 1 and 2 may become bad in the sense that it would have been better for White not
to play here at all

2
1

Diagram 1

The reasoning behind saying that is based on Diagrams 2 and 3. With a further play (triangle stone) in Diagram 2,
Black can get a good position round here, leaving White no adequate way to follow up. If the order of plays were
as in Diagram 3, wed have the same result. Most players with some experience recognise plays such as White 1
there as counter-productive. A more subtle approach is required in areas where your opponent is already strong.
Trying to develop the white stone by playing now at A will not look good when Black plays B. White is defending
without making any territory, so far; Black is attacking and gaining influence.

Diagram 2

2
1
B

Diagram 3

The loss involved in Diagram 2 or 3 should become clear when Whites other chances round here are made explicit
(Diagrams 4 and 5). These are better ways for White to become established, on whichever side makes more sense
at the time

1 2
3
4
Diagram 4

Diagram 5

4
2 1
3

Argument by transposing the order of moves has the most distinguished possible history, within the Japanese Go
tradition: it is attributed to Honinbo Dosaku (1645-1702), fourth head of the Honinbo house, Meijin, and Go Saint.
In a European foundation fable our hero, of known ancestry but unable to call on it for much in the way of practical
help, slays giants with his trusty sword and native wit. Giants, as everyone knows, are graceless creatures distinguished by aggression and greed. Dosaku finished off the giant Gluttony, who liked at all cost to capture the
opponents stones, with the sharp sword of tewari (the analysis technique we were just discussing); and his brother
Ferocity by the nimble dodging known as professional level as amashi, leaving the attacking plays as poor exchanges for the defensive replies. (For expansion on these matters, see Part II of the Yutopian book Killer of Go by
Sakata.) With these and other theoretical developments, Japanese Go was set on a course that diverged significantly
from the knife-edge balance of the ancient Chinese style.
Or at least so the story goes. It must have been recognised early on that tewari couldnt be applied indiscriminately.
Heres a worthwhile example.

Diagram 6

5
3

B A

4
1 2

This is a possible way to deal with the Chinese-style formation that appears in so many amateur games. White 1 for
Black 2 may be a loss; but after 3 Black feels inclined to play at 4, leaving White a free hand to play 5 (could be at
A also, even possibly at B) to get established in the middle of Blacks framework. If Black instead follows White
around, playing 4 as in Diagram 7, it seems as if White builds a base almost painlessly.

7 5 6
4
Diagram 7

Now if we are allowed to permute the plays at will, Diagram 8 would seem to tell us that Black has erred here.
White 1 is fair enough, Black 2 looks like a beginners cautious response why not play to the left of 3 for a proper
enclosure? Then White 3 for Black 4 is a definite loss, much like the white play in Diagram 3. But surely less
heinous? Isnt it the case that Black has played worse here, on balance?

2
3 4

Diagram 8

The distrust in tewari, as applied without discrimination by the inexpert, can be seen as stemming from a rejection
of the hidden assumption in the argument just presented. Yes, you can identify poor exchanges when you see them.
No, you cant set them off against each other. There is no just book-keeping technique of exchanges, no secure way
of cancelling them off algebraically. If you are skilful enough, you can apply this kind of analysis; if not, unfortunately, you end up proving far too much, to nobodys satisfaction but your own. It may be the condition of amateur
Go that both sides play bad moves, often in close succession. But there is no guarantee that an exchange that is
intuitively a loss can always be attributed a definite negative points value. That assumes, for example, that you can
count in the middle of a fight, as well as at momentary pauses.
Logically speaking, these values clearly do exist, at least if we are dealing with concrete whole board positions; but
why should we be able to gain access to them? This is a very interesting if perhaps fruitless point. Games of Go can
be played to an end and a definite score. A mistake costs a precise number of points against best play. What that
number is can be seen as an atomic fact about Go. We are probably not going to be able to acquire certain knowledge of these. (Cue for some talk about the early philosophy of Wittgenstein?) To be more accurate, outside
endgame positions we can only hypothesise. Professional players often talk about exchanges that are a loss, but of
less than a point. This defies pinning down, but seems to serve them well enough (oh good, chance to drag in the
later Wittgenstein too).

Lets try to get some handle on this problem with a variant. What goes on in Diagram 9 simply interpolates an
exchange 3/4 in the sequence of Diagram 6.

6 3 4
1 2

5
Diagram 9

You end up with a similar position, except that White has lost the stone 3. This looks like a pure loss? Not exactly
if White wishes to call on the sequence of Diagram 10 for eye shape, there might seem to be some point. Blacks
eye shape has of course been radically improved as well. If you are one of those players who never mentally allow
the opponent any safe territory or live group, this will be felt as a disadvantage. In truth it would adversely affect
Whites chances of fighting on the right side. In addition White has suffered a loss of potential around A.

1
3
Diagram 10

Here is a position in which I recently saw a 9 dan pro apply tewari.

Diagram 11

1
2

4
5

A strong amateur suggested White 2 in Diagram 11. The pro immediately laid out the rest of the sequence. Then
by re-ordering as in Diagram 12, he identified White 4 as a slack play: it must be at A, then Black will play at B for
a less-developed position. Here at least, when just one side is accused of a poor exchange, it seems tewari is trusted.

Diagram 12

5 A
4

1
2
3

Charles Matthews 2001