Life and Death under Chinese Rules

Consider the following board position under Chinese rules with a 5.5 point komi

  • Which stones are alive or dead?
  • Who will win this game and by how much?
  • What is the best line of play?

The purpose of this example is to explore how life and death is handled under Chinese rules. Pause here to consider the above questions if you wish to treat this example as a problem, before reading on into the rest of the discussion.

Discussion

Note that white’s group in the top-left forms a “bent four in the corner”, which is typically dead. However, the presence of an “unremovable ko threat” (marked by “a” in the next diagram) changes the situation. Consider what might happen if black attempts to capture white in the top-left, by beginning with…

White’s threat in response to the ko poses a choice to black…

If black responds to ko threat…

White will live in the top-left, since black does not have any ko threats to play (so move 9 by black can be a pass or arbitrary). While the players have traded captures, breaking up the sekis in the process, black is now losing by 0.5 points.

Alternatively, if black ignores the ko threat…

White will capture black’s bottom-right group in exchange for black capturing white’s top-left group. Yet again this exchange is not in black’s favor, since black will now lose by 2.5 points.

However, instead of playing out either of the two options above, black can simply pass (at the initial position shown in the first diagram) and accept that both the top-left and bottom-right positions are seki. With all of the stones being alive, black wins by 0.5 points (after applying the 5.5 point komi). White cannot force anything to happen, except to self-atari and increase black’s margin of victory.

Thus, it would be an absolute blunder for black to attempt to capture white’s stones in the top-left. While it is in fact true that black can claim that white’s stones in either the top-left or bottom-right could be captured, either claim would only come at the cost of sacrificing black’s stones in the opposite corner, which white would happily play out since it also flips the outcome of the game in favor of white.

Another Example

Consider another very similar example, where black is now initially losing by 1.5 points (if black were to accept all stones being alive in two sekis as is).

In this game, black should play out the capture of the bent four in the corner, while sacrificing the stones in the bottom-right (black move 5 at 1)…

After this exchange, black is now winning by 1.5 points, making the sacrifice worthwhile.

Taken together, these two examples illustrate how life and death under Chinese rules depend not only on just the local position, but also whole board strategic considerations.

Under Japanese Rules

Note that these examples are handled quite differently under Japanese rules, where the bent four in the corner would be ruled dead, even in spite of the presence of an unremovable ko threat. An older version of the Japanese actually explicitly stated that bent four in the corner was simply dead. In the more recent versions, it follows as a consequence of the life and death resolution procedure, which essentially considers the locally optimal play while disallowing any ko threats besides passing.

It should also be noted that under the modern Japanese rules, it is not automatic that bent four in the corner is always dead either. Here is an example where two interacting bent fours in the corner produces a seki: The undead bent fours in the corner

27 Likes

WOW :open_mouth: Officially nominating yebellz for ‘Awesomost Go Post of the Year’.

9 Likes

i second.

5 Likes

Third.

4 Likes

Very well explained!

3 Likes

Thank you very much!!

4 Likes

Now that this is up again, could you explain a little bit more how this is handled under the J rule? especially the difference between the “old” rule and the “more recent” version?

2 Likes

Sweet thread!! :slight_smile:

2 Likes

Well, I’m not really an expert on the Japanese rules, or any other rules, for that matter. However, I’ll try my best to explain my (potentially flawed) understanding.

It seems that there have been two official written versions of the Japanese rules: the first written version in 1949 and a revision in 1989. Prior to 1949, a formal written version did not exist, and the rules existed only as commonly understood conventions. Various rules disputes have likely played a significant role in initiating the codification in 1949 and revision in 1989.

I can’t find the full text of the 1949 rules, but based on various secondary sources (such as 1, 2, 3), it seems that this earlier version of the rules manually listed out the rulings for various special cases (including the bent four in the corner). This has been criticized for seeming arbitrary and ad hoc. Such an approach is also perhaps doomed to be incomplete, since one might find new positions that are not covered by a necessarily finite list. For example, see this related position, which is likely not covered in the special cases listed by the 1949 rules, but would be handled similarly to the bent four in the corner by the 1989 rules.

Thus, in the 1989 version of rules, the manual listing of special rulings was replaced with newly formulated rules designed to handle these special cases. A key provision is item 2 in the 7th article:

In the confirmation of life and death after the game stops in Article 9, recapturing in the same ko is prohibited. A player whose stone has been captured in a ko may, however, capture in that ko again after passing once for that particular ko capture.

Basically, this means that during the confirmation phase (consideration of hypothetical, ideal play) for determining live/dead stones, players cannot retake a ko by playing a ko threat elsewhere, and hence only passing is a valid ko threat. This rule has the implication that the bent four in the corner is dead (even with the existence of other unremovable ko threats), since black can start a ko fight in the confirmation phase and white’s attempts to play ko threats elsewhere do not allow white to retake the ko.

Ultimately, the Japanese rules can lead to very complex and difficult to understand determinations of life and death. This has lead many authors to write volumes that attempt to explain and/or criticize the rules and make many efforts to “fix” (by simplifying/clarifying) the rules while retaining the spirit.

The handling of moonshine life and anti-seki are also interesting quirks (some might say “flaws”) of the Japanese rules.

7 Likes

this is so informative i probably have to spend a day or so reading it. but i want to thank you for putting so much effort into it. this is almost like an academic paper, if not an actual book.

great job.

4 Likes

As discussed in this LifeIn19x19 thread, a similar case (a bent-4-corner seki + unremovable ko threat) emerged in the training games of the Golaxy bot:

1 Like

Unfortunately, the website linked above seems to be offline. However, I found that the original position (although without the detailed analysis) can be found via Sensei’s library here: Strange Sekis at Sensei's Library

Here is the pattern, which is seki under all rulesets:

image

This is seki in a manner analogous to hanezeki, in the sense that whoever plays another move in the local position will surprisingly start a capturing race that they lose. In other words, you can only win the capturing race if your opponent plays first (and hence no one should want to start it).

3 Likes

Wow! These are some strange positions indeed. The more I read about rulesets, the more I start to think that Chinese makes more sense.

8 Likes

Could always do with some more examples in @yebellz 's nice threads :slight_smile:

2 Likes

Yebellz’s

According to the poll in HWYDR on 24th March, the proper genitive is Yebellz’z ^^

(and the threads are Yebelzi or Yebellois~)

4 Likes

I think you can let me away with this one. I’ll probably put an apostrophe and an s when tagging :slight_smile: It just looks really clunky any other way - imagining some weird hybrid emoji speak sayin the :female_detective:'s :mag_right: or the :male_detective:'s :mag: :slight_smile:

I suppose the apostrophe z could work.

1 Like

Don’t worry, it’s just a laugh.

I genuinely do like to use Yebellz’ though.

2 Likes

Don’t I get a big say in this matter?

Why not yebel’z (yes, an “L” is intentionally dropped to trigger pedants and linguists, alike)

And how about “yebellzonian” for the adjective form?

5 Likes

So you’re shifting from yebellious?

Language is evolving!

3 Likes

Quick! Someone call the OED!

2 Likes