Repeated-Pass Rule

  Difficulty: Advanced   Keywords: Rules

This version of Spight rules is without their blind spots.[4]

  • You may not confront your opponent with a position he once was confronted with if no pass intervened.[0]
  • If you pass in the same position as once before in the current phase, it ends.
  • If a phase ends that did not charge[1] passes and you passed more often than your opponent after the first of the two passes causing its end, we regret having stopped your progress and assign you the win.
Foolish defense?  

Area scoring (with or without prisoners). 30 stones each. No prisoners. Black to move.

No matter if passes are charged or not, Spight rules deny White her cyclic defense, letting Black capture the left side and win (Black plays 1-1, she captures, he passes, she connects, . . . she may not capture his second 1-1).

Under this version, Black will only win if passes are not charged. If they are, White's defense is not foolish and Black has to be satisfied with a draw (three cyclic attacks earn him three dame, after which he switches to 2-1 and captures her throw-in[2], followed by three passes).

Only cycles without passes prevent progress and therefore have to be banned. To ban so-called odd cycles is odd.[3]

After Spight rules spoiled her defense, watch them spoil his attack.

Possible 3-for-5 cycle  

Three intersections punched out and two adjacency links added (both diagonals in the left-end square, ignoring their intersection). Scoring as above. Six black prisoners. Black to move.

Black ataris, she captures, and he passes. Instead to connect her two single stones and to be indiscriminately punished by Spight rules as in the first example, she fills liberties on the left side while he captures twice. After her capture and his snapback on the left side, she rewinds the right side while he passes. His third pass repeats his first, ending the phase.

If passes are charged, both versions let White win.

If passes are not charged, however, only this version compensates Black for prematurely stopping him. Spight rules stop him without regret and let White win (typically after another, charging phase). Note that she ironically has to choose to be more foolish because Spight rules unreasonably distinguish cycles with exactly one pass from those with more.

So, comparable to the blind spots in your visual system, Spight rules have one in the charged case and another in the uncharged. Intelligent design? Not really, even if both normally work. The one is an evolutionary patchwork and the other an oversimplification.

The repeated-pass rule is better, if not perfect. Despite being slightly more complex (regret clause) and a little more permissive about delaying the end (e.g. 2-for-1 may run twice), it bans less and always gives a sound result.

Robert Pauli


[0] RP: Some remarks:

  • Since the initial positon (to which suicide could return) is not caused by anybody, it can not be . . . with a position again if . . .
  • The position of once prevents it to mean once only.
  • Of all his former confrontations with the position in question, of course, only the latest matters.

Some substitutes (a move either is a play or a pass):

  • Ignoring what happened before the last pass, your play may not confront your opponent with a position he already was confronted with.
  • If two confrontations of your opponent with the same position are not separated by at least one pass, heck, you lost.
  • An even number of at least two consecutive plays may not recreate the position they started from.

[1] RP: Passes are charged if, effectively, who passes has to pay one prisoner, and who did not start or end half of that. Should you prefer not to use prisoners, use the same pass stones in phases that do not charge passes and score them like stones on the board.

[2] RP: If I may call it throw-in.


[3] Bill: If, with Ing, a pass is a play, the cycle that is prevented is an even cycle. To put it another way, Spight rules in this case act as a kind of situational superko rule, as White's second capture of the Black 1-1 stone would repeat a situation, not just a position. Other situations might be allowed, such as White's first capture to repeat the initial position, with Black to play. AGA rules prohibit that capture, right?

RP: This improved version of your rule, Bill, indeed is situational.

Bill: And?

[4] Bill: Apparently you think that Black should not win in this position after Black makes a throw-in ko. But where is your case? You are begging the question.

RP: Apparently Black does not throw a stone into a white eye to create a ko in the first example, if that is the example you are referring to, and apparently my rule does let Black win in both examples if passes are not charged, contrary to what your statement suggests. Where's my case? Didn't I explain that? Your rule makes no distinction where my does. Your rule is wrong in banning a cycle that neither is foolish nor stalling (first example, passes charged), and your rule also is wrong in not punishing the fool in a foolish cycle (second example, passes not charged).

PJT: Robert, I get the impression you may be misunderstanding Bill: I take him to be saying that you have not made it clear why you think it is better to make a rule that stops Black winning. On the other hand I think you may be trying to make clear what the consequences of your rules are rather than what they ought to be. My feeling is that people arguing about the rules often feel that a certain result is “fair” or “unfair” without saying why, when really one can just accept whatever rules there are and see what follows. I am more convinced by arguments about clarity, workability, simplicity and parsimony; e.g. why have rules for bent four in the corner or triple ko if you can make them follow from more basic rules. (I also wonder if you are taking “apparently” to mean “evidently” rather than “it would seem that”, a pitfall which often seems to catch out German speakers.)

Bill: I did misunderstand Robert's Foolish Cycle idea, thinking that he meant that some rules might charge one point for a pass with area scoring. It seems that all he was talking about was the loss of a point by a player for not passing, as is the case with such positions under territory scoring. AFACT, there are no foolish cycles under area scoring. If I have misunderstood, I will be glad to be set right. :)

PJT: Actually I though that RP had misunderstood you, Bill, not vice versa! (“Robert,” added to my previous remark to clarify.) But I stand by my feeling that people too often take it as given that a result of a version of a rule is unsatisfactory.

Bill: Oh, I knew your remark was directed at Robert. I was just fessing up. :)

Bill: Let's look at the play on the first board with area scoring.

Foolish defense? Not!  

B3 = pass, W4 @ 1, W6 captures
W6 repeats the original board situation. By a situational superko rule it is banned. Of course, it is banned by a positional superko rule, as well. Under area scoring passes cost nothing, so White's attempted defense, W4, is not foolish, simply ineffective.

Spight rules allow W6, because Black's pass lifted the superko ban for that position. Play could thus continue as follows. B7 @ 1, allowed because of the pass. But now W8 @ 2 is not allowed, as it would repeat the position after Black's pass. Is that a good thing or not?

Assume that all the dame are filled and the Black group on the right has only two one point eyes left. (Black could fill a dame or play inside his own territory instead of passing, to reach such a board.) Now if White could repeat the position where Black passed, Black could only pass again with safety. That would end play with the ko mouths unfilled, and White alive perforce. This position is similar to Moonshine Life. Over time, opinions about whether Moonshine Life is alive have differed, and the lore is that the original "ruling" was ambiguous. However, today all rules in wide use that I know of prevent Moonshine Life. The AGA rules do, in rare cases, permit play to end with a direct ko mouth unfilled. However, such cases do not even have the excuse that the player who benefits from that condition has an unlimited supply of ko threats, as Moonshine Life does.

Now, if you are of the opinion that play should be allowed to end in such positions, you are welcome to that opinion. But that does not mean that rules that prevent that result have blind spots. In fact, the main reason for formulating Spight rules was to prevent Moonshine Life and similar positions, such as this one, from ending with an open mouth in a direct ko, because play ended when the ko rule forced a player to pass and then his opponent passed, too. That's not a blind spot, just the opposite.


Do uncaptured pass stones make sense?

Yes, they do (as already explained in my footnote). Let the game consist of two phases, the first not charging passes and the second charging them. To let area scoring and territory scoring meet, each phase has to balance colors, and pass stones have to be treated as follows. Territory scoring has to ignore the pass stones of the first phase and treat those of the second as if captured, whereas area scoring has to ignore the pass stones of the second phase and treat those of the first as if on the board.

So, ignoring the stop for the moment, if the two-for-three cycle (driving player trades two prisoners for three) of my first example runs in the first phase (non-charging), the territory scorer will watch white prisoners increasingly outnumber black prisoners, whereas the area scorer will watch (uncaptured) black pass stones flood the, say, board extension. If that cycle, however, runs in the second phase (charging), prisoners balance out each other and the board extension remains unchanged.

Should moonshine life never survive?

The argument seems to be that it feels strange to let it survive only because the attacking player happens to run out of plays. However, if the number of intersections is limited, why not the number of plays? But since both rules in question never let the moonshine life survive, we do not have to discuss this here.

Should two for three never survive?

Spight rules won't let it. The argument (above) seems to be that since it is similar to moonshine life after the pass (if Black wastes three points), it has to share the same fate. But how can they be similar if my rule distinguishes them? Any sound definition of death will also agree with that distinction because it lets the owner of the claimed to be dead stones move first in the test. Two for three has to stay on the board for the same reason as molasses ko does.

But this isn't about special positions. By denying White to stay in the cycle, Spight rules violate a sound principle: Do not ban cycles (of an even length) that include at least one pass. Why not? Because they do not stop progress. If someone can pass, he also can tenuki. In this case, Black only can complain about being slowed down, nothing else. White is confined to a few intersection, Black is not. To punish White further is unreasonable, if not ridiculous.

Doesn't my rule also stop the two-for-three cycle?

Yes, but do not confuse stopping or aborting the cycle with banning it. Banning it, is not to allow the cycle to complete, leading to a ko fight. Stopping it, is to let the (charging) phase end as soon someone's repeated pass signals that he is done (with tenukis). Aborting it, is to spare the opponent of the fool the trouble of going thru the foolish cycle over and over again. This only happens in a non-charging phase and is merely a short cut.

Note that in this case not even the stop (let alone the ban) is justified! Why not? Because there is progress (in the lid and the extension zone). If he who benefits will not stay in the cycle forever, which is understood, there is no need for the rules to interfere at all! Therefore the regret clause has to compensate for the unjustified stop. Without that clause, Spight rules instead let the fool lose by banning him, which would be okay if they would not unreasonably only ban cycles with exactly one pass, letting the fool slip away in my second example and violating another sound principle: the fool (in the foolish cycle) always has to lose.

Shouldn't progress be mutual?

Premature end?  

Without the marked stones, this is the first example slightly modified. Three white stones were moved down one row, below the marked black stones. Though all four outcomes stay the same, the draw is different (remember, it only happens under this rule version and with passes charged).

Instead to exploit his cyclic attack to fill three dame, Black now spoils three white territories by throwing a stone into each, and to prevent their capture, he does not make seki (on 2-1) but now continues his cyclic attack (on 1-1) until his repeated pass ends the phase, reaching the position shown (which has to be balanced with two white prisoners if you prefer territory scoring).

Did Black finish? Of course he did. He can't seriously compare this to a moonshine life that survived because he ran out of plays. Obviously he could have stuffed two more stones into his territory without getting any closer to a capture, and no other number would have changed that.

Did White finish? No, she did not. She would have loved to capture the stranded three black stones, but Black didn't let her. Should Black be banned? Well, she has a point in complaining that Black continued his cyclic attack for no other reason than to keep her busy (remember, passes are charged, the cycle as such earns Black nothing). On the other hand, you can't have it all. If the repeated-pass rule allows her a cyclic defense, which Spight rules deny her here, she has to pay the price for taking that option and skip her clean-up.

Repeated-Pass Rule last edited by RobertPauli on November 21, 2019 - 15:07
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Login / Prefs
Sensei's Library