What is good shape?

British Go Journal No. 62. July 1984. Page 24.

Francis Roads

A game of go is a living organism. You can never fully understand one part or aspect of it without taking the whole into account - and the whole is usually beyond human understanding.

This is one reason why you never read an attempt to present the principles of Go in a systematic fashion - as one might write a mathematics or physics text book. Japanese authors are particularly prone to presenting information in disconnected gobbets, leaving the reader to make the necessary connections. Nowhere will you see this approach more clearly than in the book "Go Proverbs Illustrated" by Kensaku Segoe. Many of the proverbs are concerned with "good shape", the translation of the Japanese "katachi". A better book on this important subject I have yet to see. But nowhere in the book or elsewhere for that matter - have I seen a definition of what constitutes "good shape".

The term is usually employed where there are two or more moves that achieve a given objective. The goad shape move is the one that not only achieves the aim, but gains some other advantage, which may, however, be small and intangible at the time. The player with an understanding of good shape has a short cut to finding the best move in many situations. In a game free of gross blunders the accumulation of small advantages from good shape will decide the outcome.

Without further ado, therefore, here are my criteria for a move to qualify as good shape:

  1. It maximises liberties
  2. It maximises eye-making potential
  3. It keeps options open
  4. It influences as much of the board as possible
  5. It denies the opponent good shape

Of course not all these criteria must apply for a move to be good shape. Indeed sometimes they may mutually conflict. However they should prove helpful in finding the right move.

Lets look at these criteria in the light of some examples.

Diagram 1



Diagram 2



Diagram 3



Dia 1 shows the most basic type of good and bad shape. Black, for same tactical reason, needs to connect his stones out to the right. To play at A produces the notorious "empty triangle", shown in Dia 2. A play at B produces the good shape of Dia. 3. Why is the empty triangle bad? First, it does not maximise liberties. A three stone group can aspire to eight, but the empty triangle has only seven.

Diagram 4




Second, if you look at the typical eye-shape in Dia 4, you will see that there is only one way for Dia 2 to become part of an eye. Dia. 3 however has two ways to do so.

Third, in Dia 3 black has the option of sacrificing the two stones separately from the one. In Dia 2 all three must be saved or lost.

Fourth, the influence of the stones in Dia 3 extends one line further to the right. Finally, criterion 5 does not apply in this case.

"Isnt this all rather clinical", I hear you saying. "Does it make so much difference whether a group has seven or eight liberties, or whether there are one or two theoretical eye shapes".

Far from it - these are precisely the small intangibles referred to earlier. Here is a less theoretical example.

Diagram 5








Diagram 6








Diagram 7








After the well known joseki in Dia 5 black usually needs to protect his cutting point at C if his stones are to exert sufficient influence to compensate for the points in the corner. Which do you prefer? Dia. 6 or Dia 7?

Well done - Dia 7 is of course better. There is no empty triangle. There is better eye-making potential, and, once again, the influence extends further to the right. In this case there is the added advantage that whites endgame play at D is gote.

But there is a price to be paid for this goad shape. A strong ko-threat exists for white at E, which was not available in Dia 6. Nonetheless this is not a good enough reason for choosing Dia 6 in the opening.

Diagram 8









Diagram 9









Bad shape.
Diagram 10









A bamboo joint.

Dia 8 will be familiar to readers who like to play on the three-three point. As in Dia 5, black must protect his stones from a cut at F to secure his outside influence.

Only a beginner would connect directly at F, leading to Dia 9, which directly violates criteria 1-4, So how about G, leading to Dia 10? This is the well-known 'bamboo joint', and an improvement on Dia 9 - better eye potential, slightly more influence, and the ability, if necessary, to sacrifice two tail-end stones.

If we hypothesise the exchange white H, black I, then blacks formation still has one more liberty than in Dia 9.

The above are typical of the thought processes in identifying good shape.

But lets keep looking.

Diagram 11








Diagram 12








What about J in Dia 8, leading to Dia 11? This prevents the push through at K, as black answers at L. But white has a good forcing move at M, which black can only answer by K or N, in either case making the dread you-know-what.

Another point to note: where the opponent has a forcing move which immediately turns good shape into bad, the shape is suspect to start with.

Better than all the moves examined so far is O, leading to Dia 12. But there are still some awkward forcing moves against it (see if you can find them).

Diagram 13









Best of all.
Diagram 14









A ladder.

Best of all, in fact, is playing at P in Dia 8, leading to Dia 13. White cannot push and cut because he is caught in a ladder, Dia 14.
[Ed: and blacks stones have maximum flexibility and - important in this position - influence.]

Until now weve seen little of criterion 5, so lets look at Dia 15.

Diagram 15








Where should black play? White is hoping black will fear the cut at Q and take time out to protect it (how would you protect it - connect at Q or ...?).

Diagram 16








Instead, however, black has a really superb attacking move at R. This is the "eye-stealing tesuji" of Go Proverbs Illustrated. White was desperately hoping to play on this vital point himself, more or less guaranteeing himself an eye by playing S or down along the edge. But once a black stone appears there not only does his eye vanish, but he cannot connect his stones without making an empty triangle. Conversely, therefore, black R is regarded as good shape for black.

[Start]


This article is from the British Go Journal Issue 62
which is one of a series of back issues now available on the web.

Last updated Thu May 04 2017.
If you have any comments, please email the webmaster on web-master AT britgo DOT org.