scoring eyes in seki question

[Dusk Eagle]

Yes, the original position is just a seki. No points for either side.

[Dusk Eagle]

In practice, you don't have to fill all the dame

For the sake of simplicity, let us stick to the two-sided dame.

Practice differs for real world go played under the Japanese 1989 Rules (dame filling can affect the score, if the rules are applied strictly), real world go played under the World Amateur Go Championship Rules (dame filling is required), real world go played under variants of verbal Japanese rules (usually, dame filling is a custom, but the exact timing and nature varies; at the end of counting and before announcing the score, the dame should be filled), online go with Japanese server rules allowing unfilled dame (dame filling is voluntary, but - in case of especially KGS - is recommended so that server scoring mistakes do not occur and affect the score), online go with Japanese server rules requiring dame filling.

Tournament rules might prescribe alternate dame filling.

Conclusion: depending on the place of play, the used version of Japanese rules and any used server, filling two-sided dame is or is not mandatory and is or is not relevant for the score.

I agree with what you've said I think. The point I was trying to make is that in the cases where the dame aren't filled before passing and it does somehow affect the score, (e.g. someone insists on a "any adjacent dame" interpretation of seki,) the game can be resumed and all of the dame can be filled.

[RobertJasiek]

Resumption is possible AFTER game-stopping passing and BEFORE the game end:)

[Mef]

So then, back to the original question..

There's never a case where some set of eyes in seki would count as points? In practice, or theory?

I think in practice (irregardless of the actual rules), we would count the eyes in diagram 15 above, and not even call that a seki, right?

But in my original position, its a true seki and none of the eyes are counted then?

Strictly speaking, you should not count the points unless someone is willing to fill the dame. It is incredibly rare, but there are situations where groups with two eyes make a seki. Imagine this --There is a ko somewhere else on the board white can start, but white's threat with the atari is not large enough to win. If black plays the final dame, black will give white a threat (because now the corner threat is 2 points larger than if white filled the dame!). Black will not play because it gives white a ko threat. White on the other hand stands to gain 1 point by not playing the dame, because he loses 2 point in the seki to black's 3. If neither player can play the dame without taking a loss, it is a seki.

[RobertJasiek]

Interesting... can you also construct such a position?

[Mef]

Interesting... can you also construct such a position?

I have seen them constructed in the past (so I know at least one exists)...I can try to work on either finding that or making a new one when I get the time.

[Mef]

Interesting... can you also construct such a position?

I have seen them constructed in the past (so I know at least one exists)...I can try to work on either finding that or making a new one when I get the time.

Ok, I did some digging and it looks like this has actually come up on L19 before!

A position for your consideration:

$$c Unfillable Dame? \n(Japanese Rules, No prisoners, 5.5 Komi).
$$ ---------------------
$$ | . . O . . O X . X |
$$ | O O O O O O X X O |
$$ | X X X X X O O X . |
$$ | . . . . X O T X . |
$$ | . X . . X O O X X |
$$ | X . X X X X O O T |
$$ | X X O O O X X O O |
$$ | O O . . O X . X O |
$$ | . . O O O X X X O |
$$ ---------------------

Click Here To Show Diagram Code

[go]$$c Unfillable Dame? \n(Japanese Rules, No prisoners, 5.5 Komi).
$$ ---------------------
$$ | . . O . . O X . X |
$$ | O O O O O O X X O |
$$ | X X X X X O O X . |
$$ | . . . . X O T X . |
$$ | . X . . X O O X X |
$$ | X . X X X X O O T |
$$ | X X O O O X X O O |
$$ | O O . . O X . X O |
$$ | . . O O O X X X O |
$$ ---------------------[/go]

[RobertJasiek]

It behaves differently under different Japanese rulesets (at least J2003 (dame create in-seki), WAGC (dame filling is required) and verbal J (dame filling is not required) have three different behaviours). The position is interesting, but, for each ruleset, I need time to do a proper analysis of variations and scoring after each. Under J1989, there is also the ambiguity of the black ko stone's status if the dame are filled and the position is scored then; under J2003, there is no ambiguity.

[Bill Spight]

Interesting... can you also construct such a position?

I have seen them constructed in the past (so I know at least one exists)...I can try to work on either finding that or making a new one when I get the time.

Ok, I did some digging and it looks like this has actually come up on L19 before!

A position for your consideration:

$$c Unfillable Dame? \n(Japanese Rules, No prisoners, 5.5 Komi).
$$ ---------------------
$$ | . . O . . O X . X |
$$ | O O O O O O X X O |
$$ | X X X X X O O X . |
$$ | . . . . X O T X . |
$$ | . X . . X O O X X |
$$ | X . X X X X O O T |
$$ | X X O O O X X O O |
$$ | O O . . O X . X O |
$$ | . . O O O X X X O |
$$ ---------------------

Click Here To Show Diagram Code

[go]$$c Unfillable Dame? \n(Japanese Rules, No prisoners, 5.5 Komi).
$$ ---------------------
$$ | . . O . . O X . X |
$$ | O O O O O O X X O |
$$ | X X X X X O O X . |
$$ | . . . . X O T X . |
$$ | . X . . X O O X X |
$$ | X . X X X X O O T |
$$ | X X O O O X X O O |
$$ | O O . . O X . X O |
$$ | . . O O O X X X O |
$$ ---------------------[/go]

Sakai's Anomaly. Under J89 rules White to play passes and Black cannot afford to fill a dame. White wins by 0.5, because of the weird seki, which exists only under J89 rules.

Play goes smoothly under Double Button Go, where the players hand over pass stones, except when the same player makes the last pass and the first pass. Here is an SGF file for Double Button Go with two consecutive passes to end play.

(;SZ[9]FF[4]AP[GOWrite:2.2.21]ST[2]CA[ISO8859-1]GM[1]C[White to play.]FG[259:]AB[dc][fg][he][cc][ee][ec][gg][ga][ie][bc][ac][ia][be][fi][gi][hi][hc][df][ef][fh][ed][ff][hh][af][hd][cf][hb][gb][bg][ag]GN[ ]KM[5.5]PM[2]PW[ ]AW[eg][ge][cg][ca][fc][dg][fe][ig][ii][fa][hg][ei][gc][ci][di][bb][cb][ih][ab][gf][hf][fd][ah][ib][fb][bh][db][eh][eb]PB[ ]
;C[*** White passes, handing over a pass stone.]W[tt]
(
;B[gd]
(
;W[if]
;B[ic]
;C[*** White passes, handing over another pass stone.]W[tt]
;C[*** Black passes, handing over a pass stone. White has made 1 more pass than Black, so Black wins by 0.5.]B[tt]TW[aa:ba][ai:bi][ch:dh][da:ea]TB[ad:dd][ae][bf][ce:de][gh][ha][ib][id]
)
(
;C[*** White passes, handing over another pass stone.]W[tt]
;B[if]
;W[ha]
(
;C[*** Black removes a ko threat.]B[ae]
;W[ic]
;B[ia]
;C[*** White removes a ko threat.]W[bi]
;B[id]
;C[*** White passes, handing over a third pass stone.]W[tt]
;C[*** Black passes, handing over a pass stone. White made 2 more passes than Black. Result\: Black +2.5.]B[tt]TW[aa:ba][ai][ch:dh][da:ea]TB[ad:dd][bf][ce:de][gh][ha][ib:ic]
)
(
;C[*** Black passes, handing over a pass stone.]B[tt]
;W[ic]
;B[ia]
;W[ae]
(
;B[ad]
;W[ha]
;B[ch]
(
;W[id]
;B[bi]
;C[*** White passes, handing over a pass stone.]W[tt]
;C[*** Black passes, handing over another pass stone. White made one more pass than Black, for a net cost of 1 pt. So Black wins by 0.5.]B[tt]TW[aa:ba][da:ea][ga][gb:hb][gd:hd][hc][he:ie][ia][if]TB[ae][ah:bh][ai][bd:dd][bf][ce:de][cg:eg][ci:ei][dh:eh][gh]
)
(
;W[bi]
;B[ia]
;C[*** White passes, handing over a third pass stone.]W[tt]
;B[id]
;C[*** White passes, handing over a fourth pass stone.]W[tt]
;C[*** Black passes, handing over a pass stone. White has made 2 more passes than Black. Black wins by 2.5.]B[tt]TW[aa:ba][ai][ch:dh][da:ea]TB[ae][bd:dd][bf][ce:de][gh][ha][ib:ic]
)

)
(
;B[id]
;W[bf]
;B[ad]
;W[af]
;C[*** Black passes, handing over a pass stone.]B[tt]
;C[*** White passes, but, because White made the first pass as well as the last, White does not hand over a pass stone. Both players have handed over the same number of pass stones. Result\: 7.5 for White.]TW[aa:ba][ag:bg][ai:bi][ch:dh][da:ea]W[tt]TB[bd:dd][ce:de][gh][ha][ib:ic]
)

)

)

(
;C[*** White removes a ko threat.]W[bi]
;B[if]
;W[ha]
;C[*** Black removes a ko threat.]B[ae]
;W[ic]
;B[ia]
;C[*** White passes, handing over another pass stone.]W[tt]
;B[id]
;C[*** White passes, handing over a third pass stone.]W[tt]
;C[*** Black passes, handing over a pass stone. White has made 2 more passes than Black. So Black wins by 2.5.]B[tt]TW[aa:ba][ai][ch:dh][da:ea]TB[ad:dd][bf][ce:de][gh][ha][ib:ic]
)

)
(
;B[ic]
;W[gd]
;B[if]
;C[*** White passes, handing over a second pass stone.]W[tt]
;C[*** Black passes, handing over a pass stone. White made one more pass than Black. Result\: Black +0.5.]B[tt]TW[aa:ba][ai:bi][ch:dh][da:ea]TB[ad:dd][ae][bf][ce:de][gh][ha][ib][id]
)

)

[ashleyco]

There's never a case where some set of eyes in seki would count as points? In practice, or theory?

[leichtloeslich]

There's never a case where some set of eyes in seki would count as points? In practice, or theory?

Not in Jap. style rules, no.

Btw, I have to date not seen a satisfactory formal definition of "eye". Currently browsing http://senseis.xmp.net/?FormalDefinitionsOfEye.

From what I've seen I'm lead to believe that there cannot be such a definition. However, if there ever were such a definition, I would prefer that it complies with the rule "two eyes alive" in the sense that a group with 2 "real eyes" (however that's defined in the end) cannot be captured even if the owner passes continuously.

Note that this is clearly not the case in the OP.

[RobertJasiek]

I have to date not seen a satisfactory formal definition of "eye". Currently browsing http://senseis.xmp.net/?FormalDefinitionsOfEye.

It depends on your purpose of application.
- My definition works for all scoring positions for the purpose of scoring, and works (by imagining successive passes) for all positions considered as if they were scoring positions and for the purpose of scoring.
- My semeai-eye definion works for all basic semeais with strong eyes.
- If your purpose is not scoring, but is strategy, a formal definition has not be stated yet. However, it is possible along the idea of considering an intersection, or a connected set of intersections, and verifying whether the opponent can prevent the player's creation of a scoring-eye on at least one of the intersections.

From what I've seen I'm lead to believe that there cannot be such a definition.

For some purposes, there already is such a definition, see above. For other purposes, such a definition can be found, see above.

if there ever were such a definition, I would prefer that it complies with the rule "two eyes alive" in the sense that a group with 2 "real eyes" (however that's defined in the end) cannot be captured even if the owner passes continuously.

Needless to say, such makes sense only for non-seki eyes!

Presuming non-seki scoring eyes, your wish is fulfilled due to the proof of Chris Dams, which relates capturability-life to two-eye-alive:
http://home.snafu.de/jasiek/wagcmod.html
(Google for the actual proof.)

[Bill Spight]

Btw, I have to date not seen a satisfactory formal definition of "eye". Currently browsing http://senseis.xmp.net/?FormalDefinitionsOfEye.

Formal definitions are not always satisfactory. Here is one.

An eye is a connected region of empty points adjacent to two or more stones of one color and no stones of the other color. Now, that definition is unsatisfactory in at least two ways. First, two such eyes do not necessarily confer life. Second, there are regions that are considered eyes that do not fit the definition. Still, it could be perfectly serviceable in a theory of eyes.

From what I've seen I'm lead to believe that there cannot be such a definition. However, if there ever were such a definition, I would prefer that it complies with the rule "two eyes alive" in the sense that a group with 2 "real eyes" (however that's defined in the end) cannot be captured even if the owner passes continuously.

IMO you are asking for too much. For instance, there are sekis in which a false eye is necessary for life. We need definitions for different kinds of eyes and, if we define eyes by morphology, we may never be sure that we have discovered all the different types of eyes.

You may be interested in Landman's paper, in which he defines topological eyes and fractional eyes. http://library.msri.org/books/Book29/files/landman.pdf

[mitsun]

Play goes smoothly under Double Button Go, where the players hand over pass stones, except when the same player makes the last pass and the first pass. Here is an SGF file for Double Button Go with two consecutive passes to end play....

Did you leave out the simplest variation? If W starts by passing, can B respond by passing, ending the game instantly with a 1/2 point victory?

[Bill Spight]

Play goes smoothly under Double Button Go, where the players hand over pass stones, except when the same player makes the last pass and the first pass. Here is an SGF file for Double Button Go with two consecutive passes to end play....

Did you leave out the simplest variation? If W starts by passing, can B respond by passing, ending the game instantly with a 1/2 point victory?

If the game goes pass-pass, the yose ko in the top right corner is not settled. Double Button Go allows it to be played out without stopping play first.

Under Japanese '89 rules, Black does pass, but she then loses the game. Seki is defined by having dame, so that Black gets only 9 pts. and White gets only 4. Then White wins by 0.5. The White stone in the yose ko is dead, but cannot be removed because of the seki. (The Japanese '89 rules are so bad!)