What is the score under this seki situation? Go forum
13 replies. Last post: 2019-05-20Reply to this topic Return to forum
gamesorry at 2019-05-12
One of my games is in the final score counting stage: http://littlegolem.net/jsp/game/game.jsp?gid=1966165
I posted the snapshots of the two (wrong) results on L19:
It seems that there’s no way to get to the correct score on LittleGolem? (could only get either B+3.5 or W+2.5, but I think the correct result might be W+0.5?) (With Japanese rule, it seems white might need to capture the two stones and black place one more in it, so white gets 3 pts for the seki group while black has 0 pt for the seki group?)
eobllor at 2019-05-12
As far as I know, in the Japanese rules, the territory within a seki does not count, thus the result would be B+0.5 (once W has captured the two B stones).
If territories in seki do count, then the second result (W+2.5) seems correct to me, since B cannot force W to capture the two stones and therefore W is not required to do it.
Malcolm Schonfield ★ at 2019-05-12
I strongly dislike Japanese rules as they are needlessly complex. If we take Chinese, US, French or NZ rules (along with the 6.5 komi) then White wins by 2.5 points. However in LG it is unfortunately not clear what rules are being used! I suggest scoring the game as if the two stones are dead; if there is a disagreement, then play resumes, and the 2 stones are captured, leading to win for White by 0.5 points. So in this case it doesn’t change the result in an important way. Still, it’s an interesting situation; thanks for sharing. 谢谢!
It would be nice if some day we could agree on the rules being used, but I’m not sure it’s very likely to happen..!
Hope this helps.
GoNut at 2019-05-13
Under Japanese rules, territory within the seki doesn’t count, but the two black stones are considered dead and belong to White’s tally. So it’s B+0.5
Under Chinese rules (komi 6.5), with the dame pts all filled in and area counting applied, it’s W+2.5
Little Golem uses territory counting as per Japanese rules (territory enclosed + captured/dead oppoent’s stones), so it’s B+0.5.
This game is a good example of the lack of precision and illogicality of Japanese rules in certain situations: white encloses more territory in its seki group than black, so not including seki territory in the final count disadvantages white. The other point, mentioned by Malcom, is: what if there’s a dispute as to whether the 2 black stones are dead? In Chinese rules, the game is resumed and played out. In Japanese rules, there’s no such provision. (Interestingly, if black were to force white to capture the 2 stones, it would have to place an additional stone in, resulting in W+0.5 – but this is moot, since Japanese rules do not provide for resuming a game after 2 passes).
Malcolm Schonfield ★ at 2019-05-13
@GoNut : I agree LG uses territory counting, but territory counting is also used by other rulesets, eg French. I see no reference in LG that Japanese rules are used. I think it’s important to make the distinction between the ruleset and the counting method.
Do you have any references that LG uses Japanese rules?
GoNut at 2019-05-13
I get what you mean. Actually I don’t have any references. I’ve always assumed that LG’s territory scoring was based on the Japanese rule set, since that’s the rule set I’m most familiar.
Apologize for the oversight.
Carroll ★ at 2019-05-13
On the Go rules page www.littlegolem.net/jsp/games/gamedetail.jsp?gtid=go19&page=rules, it is written:
"On this server is go played according to Japanese rules. Komi is:
- Go 9x9 – 5.5 komi
- Go 13x13 – 8.5 komi
- Go 19x19 – 6.5 komi"
GoNut at 2019-05-13
Thanks, Carroll. Most helpful.
Malcolm Schonfield ★ at 2019-05-14
@Carroll: thanks! @goNut: no worries.It could perhaps be argued that the result should be B+3.5! Why? because:
a) after move 244, both players have passed,b) no territory is counted in sekis in Japanese rulesc) no proviso exists in Japanese rules to resume play after two successive passes.
See the position at move 245.
I’m not 100% sure of the above, because it’s Japanese rules and I do not pretend to fully understand them :)
However play has now resumed and I think it’s heading towards the result of W+0.5. If I understand correctly this would have been the result under formal Japanese rules, if we do not take into account the passes.The thread on life in 19x19 has some explanations of why further moves are necessary in Japanese rules.
I wonder if it is possible to get LG to switch to a more reasonable ruleset! :)
GoNut at 2019-05-14
After move 244, LG made 2 errors in counting (under Japanese rules)
1. The single point in Black’s seki territory was counted, giving an erroneous final score of B3.5;
2. The 2 black stones in White’s seki territory are obviously dead, because they do not form the boundary between Black and White’s territories and cannot be re-connected to Black’s seki group. So these should be lifted off the board and added to the White’s score.
So after move 244, under formal Japanese rules, the score should be B+0.5 (+3.5-1-2). No Japanese professional would place a black stone in to force the capture of the 2 black stones as this would disadvantage his score by 1; instead the two players would simply agree that the 2 black stones are dead.
The game has now resumed in an attempt by the players to get LG’s scorer to count correctly – so we see W sportingly capture the 2 black stones (without B placing an extra stone in to force the capture) to maintain the score at B+0.5.
It’s very difficult for AI programs to determine winning percentages accurately under Japanese rules because of various uncertainties and some arbitrary life-and-death rules (e.g. ‘bent four in a corner’), so it’s no wonder that most top programs like Alphago, Leela Zero and Fine Art use the Chinese rule set for scoring.
For over-the-board human games, for ease of counting, the AGA rule set would be a good alternative: it reconciles territory counting with Chinese rules/area scoring by introducing the concept of ‘pass stones’.
GoNut at 2019-05-14
Correction to my post above:
After move 244 and 2 passes, it’s B+0.5.
After resumption of play, @Maxwell Schonfield is right, it’s headed for W+0.5, because when white captures the 2 black stones, black has to throw another stone in, resulting in a 1-point swing to W.
The fact still remains that there’s no provision under official Japanese rules for play resumption after 2 passes. This may be problematic if the life (or death) of certain groups cannot be agreed upon by both players. In this game W shouldn’t have passed at move 244 after B passed, but should instead have proceeded to capture the 2 black stones to force a 3rd black stone to be thrown in.
gamesorry at 2019-05-20
Thanks for the confirmation, we managed to get to the W+0.5 score in the end (luckily each group ended up with one eye so the extra points cancelled out) :)
eobllor at 2019-05-20
Ah, I had missed that stone Black needs to sacrifice in the end! Nice catch.