Why not? However, how can you use a suicide for winning a
If I play a suicide move, your group will never have less
liberties than before the suicide. Actually, it will have
more liberties (my removed stones).
Sergio, just curious
Since I heard about some traditional Chinese rules (where suicide was
allowed) and started studying rule issues, my opinion was that suicide
should be allowed. There is no "reason" for it. It makes the game
simpler and richer. The rule confuses beginners. (I don't teach it.) I
think that allowing suicide might also throw a curve ball at the go
playing programs by increasing the number of possibilities.
Much like ishi-no-shita, you have to see under the stones. In the
construction of the AGA rules, this was the one issue on which Bob
High (and other more tradition minded AGA members of the rules
But the major go playing countries - including now China - do not
allow suicide so a change will await a change at the pro level. I'm
glad Ing allows suicide.
Richer: yes. Strategic variety is increased.
But why simpler? It is not a matter of having or not having a suicide
rule but about completing any definition of "play". Somehow or
otherwise it must be specified what happens in case of liberty-less
own stones during absence of liberty-less opposing stones.
"first remove opposing stones without liberty, then remove the
player's own stones without liberty"
"first remove opposing stones without liberty, then the player's own
stones are required to have a liberty".
Pretty much the same simplicity from a POV of understanding.
I have not considered it in winning a semeai. Is that possible?
I have always urged Go programmers who aren't keen to implement all
possible rulesets, to be conservative in the moves that they make, and
liberal in the moves that they anticipate from their opponents. Thus in
the case of suicide, they should not get their programs to make suicide
moves, but should get their programs to accept suicide moves from their
When I ran a computer Go tournament using Ing moves
http://www.computer-go.info/events/ing/1998/index.html , one of the
games was decided when a player made a suicide move (for no good reason,
it was a very weak program) and its opponent refused to acknowledge the
>Much like ishi-no-shita, you have to see under the stones. In the
>construction of the AGA rules, this was the one issue on which Bob
>High (and other more tradition minded AGA members of the rules
>But the major go playing countries - including now China - do not
>allow suicide so a change will await a change at the pro level. I'm
>glad Ing allows suicide.
Nick Wedd ni...@maproom.co.uk
> Since I heard about some traditional Chinese rules (where suicide
> was allowed) ...
> But the major go playing countries - including now China - do not
> allow suicide so a change will await a change at the pro level.
> I'm glad Ing allows suicide.
As you say, modern Chinese rules do not allow suicide (e.g.
and http://www.cs.cmu.edu/~wjh/go/rules/Chinese.html ).
And Ing rules are probably too recent to be considered as a
But what were these traditional Chinese rules of which you speak?
Are they described in, say, GoGoD, or somewhere else that is accessible
to an English reader? I have not heard of this before. I thought the
actual use of a suicide rule was largely a modern development, though
no doubt many players over the centuries have thought of it.
Another example, similar to the above but with more explanation,
is given at Senseis Library:
With no suicide allowed, the result is Black winning the capturing
race, killing the White stones. But with suicide, the result is seki.
I do not know of an example where the outcome is a total reversal
when suicide is allowed. That is, Black kills White when suicide
is not allowed, but White can kill Black when it is allowed, rather
than just obtain a seki. Does anyone know of such an example?
This will be a somewhat unsatisfying answer. In the 80s (one of the
moments when this issue was "hot") I heard that different areas of
China had different rules including allowing suicide. There were no
citations. (These were all unwritten rule sets.) I was told this (as
best I remember) by some of the Ing folks who favored allowing
suicide. Yang Yuchia of the Ing Foundation might have more information
The question of suicide has been under discussion since the Igo Kempo
articles of the 20's and 30's. Robinson and Olmstead picked up on it.
I think it's simpler to have a single rule for removal of stones
(first any opposing stones with no liberties and then own stones with
no liberties) rather than a rule for removal and then a prohibition on
moves which create a position where there is no opponent removal. And
- whether the rules can be shorter or not - the concept is easier for
beginners to understand without the suicide rule.
Letting a beginner play a move which results in self capture is an
effective way to teach them not to do it. I play, my stones go off,
and they you play. Oops.
But this is all irrelevant at the moment. The suicide rule is not "on
the table". We have enough to contend with in the issues of single
sided dame, seki, super ko, and the last point played.
Sure but both versions can be written in simpler or more difficult
wordings. E.g., J1989 describes it unnecessarily difficult.
Aesthetically, it is inconsistent with the normal
sequence 'white plays and captures the black group.'
Personally, I agree.
Logically, it seems arbitrary, to allow or disallow.
In a similar vein, the japanese rule that scores
no points in a seki, is aesthetically displeasing.
But, according to Chen Zuyuan, historically explained by the
development from Tang Rules to Japanese Rules:
Can you translate that verbiage into plain english?
Tell us what of it you don't understand and some explanation might be
I don't see what connection it has to japanese rules
for scoring seki.
Tang Rules were stone scoring rules but counted in Japanese style
(territory minus prisoners) modified by the group tax. During the
process of exporting the rules to Japan around the Tang era,
apparently the group tax was lost / abandoned while a new rule of no
territory in sekis emerged.
Counting territory in basic, ordinary sekis but applying the group tax
to them equals not counting that territory in sekis. So what was a
counting aspect became the seki scoring reality in Japan.
In later centuries, Japanese players forgot the earlier history and
started to believe that their reality then was the "natural" thing for
sekis while in reality natural consistency of territory everywhere was
lost due to the historical accident of confusing a counting method
with the scoring and forgetting about the group tax element of
I hadn't seen the article by Chen - WOW! What a great piece of
research on the rules.
It only addresses one of the outstanding issues - counting and - other
than an off hand comment that it was probably not allowed - it doesn't
address the topic of this thread: suicide. Nor is there much help for
the other open rules issues: super-ko, bent-4, single sided dame and
other quirks in the end game.
So a traditionalist argument based on rules history would support
counting points in a seki. Good. Secondly tradition would insist that
each side play an equal number of stones. OK. AGA rules do that and
then counts all 361 points. Chen's suggestion of not counting the last
dame point at all if Black would get it means that sometimes only 360
points are counted. That means a granularity of one in the scoring of
the game - instead of two under AGA or Chinese area rules. But at what
cost in other difficulties?
Thanks much for the citation.
No. Rather it means that currently the empirical frequency of one as
the smallest score difference is greatly increased. The theoretical
frequencies of even and odd scores under area or stone scoring is an
unsolved problem (despite my prize for its solution;) ).
Greatly increasing the frequency of a one point score difference
matters in the question of komi. The Japanese really want to use a 6
1/2 komi - only meaningful with a granularity of one. That's the
advantage of a smaller granularity. I don't think that trumps the
other issues involved but it does matter.
Once more: It is NOT the granularity that is smaller (the granularity
is the same: one) but the PRACTICAL / EMPIRICAL FREQUENCY of odd
versus even score differences on the board. So it is not an advantage
of a particular granularity but an advantage of a similar practical /
empirical frequency of odd versus even score differences on the board.
OK, thanks for your usual precision, but this is also about words and
I find "granularity" the more efficient way to talk about the
difference. A 6 1/2 komi is virtually meaningless because the
"effective" granularity (a "soft" term) of area rules is two. That
will be more understandable to more players than: a greater empirical
frequency of odd versus even score difference in area rules.
I'll try to be more precise.
What do you think is the percentage of even score differences in AGA/
On an odd board like 19x19, I presume.
Theoretical percentage: I do not have the slightest idea. My guess is:
less than 50%. So far I can prove only the lower bound that it is more
than 0% and the trivial upper bound less than 100% :)
Practical percentage: It depends by whom the games are being played
because on KGS there are about 35 times as many sekis as created in
Japanese professional games, roughly every 2nd game instead of every
70th game. I forgot the distribution of even versus odd sekis; let me
guess for the moment a relative frequency of 2:1. So on KGS roughly
every 6th might have an odd seki for an even score difference on the
board while among Japanese pro games roughly every 210th not resigned
game has it at its end. But more statistics should be done for sure.
> Theoretical percentage:
agree not 0 and not 100. ;-)
> Practical percentage: It depends by whom the games are being played because on KGS there are about 35 times as many sekis as created in
> Japanese professional games, roughly every 2nd game instead of every 70th game.
35 times? Every second game? Wow. I wonder why. Is that a good
In my tournament games I'm only a bit above the Japanese percentage
for sekis. What is it in your games?
>I forgot the distribution of even versus odd sekis; let me guess for the moment a relative frequency of 2:1.
Agree and close enough for argument. I almost said in an earlier post
(without being as precise) that for amateurs in tournament or club
games odd sekis would be less than 1%.
The KGS stats seem an anomaly.
> So on KGS roughly every 6th might have an odd seki for an even score difference on the
> board while among Japanese pro games roughly every 210th not resigned
> game has it at its end. But more statistics should be done for sure.
What might be easier (seems to me) is not a random generator but a
program which could acquire a large number of actual game records (pro
or amateur) and count the sekis.
* * *
When we are talking about international rules with the pros, they base
their opinion on the rarity of odd sekis in their games. Clearly, for
pros, there is a significant difference in the effective granularity
between the two rule types and therefore why they hold so firmly to
their different positions. It's also why finding a way to reduce the
effective granularity to be closer to one is important - no matter how
We may yet come back to the button or a decision phase or some such
device. Unaesthetic but functional.
Unfortunately there doesn't seem to be much of an ongoing discussion
at the moment.
Much smaller thinking times and much fighting lead to lots of
unreasonable fights that by accident can end in sekis at times.
>Is that a good statistic?
I think Japanese games I checked ca. 200 but the sample may have been
prejudiced. KGS: my rough estimate from my experience of playing /
> What is it in your games?
>The KGS stats seem an anomaly.
We need to check more sources...
>Unfortunately there doesn't seem to be much of an ongoing discussion
>at the moment.
You are free to write more:)
(Currently I am not: I am going to attend the European Congress for
the next 15 days.)