I’m not a native speaker of Japanese or English, but when using “dame” as a go term in Dutch, I’d also consider “dame points” to be empty intersections, where “dame points” basically means “neutral points”. The other main type of empty intersections being “territory points”.
Even in the context of “playing on dame points”, I’d imagine that “dame points” refers to the intersections where the stones are placed on, not to the stones played on those intersections.
I don’t think I would still call those intersections “dame points” once they have been occupied (but I’m not 100% sure if I’d be fully consistent in this).
In that case they can score as they agree. OGS behavior (overriding agreement by unrequested analysis and outside info) is dubious in these cases. However that is not teire anymore but an unsettled position, with an “effective move” revealed to both on the board, so more about resumption rules than scoring. As it stands W has no corner territory neither in J89 analysis (stones not alive) nor from control view (opponent first).
(Interesting position btw, particularly after some modifications so that B can still capture some W stones, but he doesn’t get to choose which…)
I was reading the rules commentary and 2003 rules of Robert Jasiek too and is this correct? My understanding of the 2003 rules is that the two stone string is capturable-2 With White losing 2 stones in the process (I’m just describing how to establish its capturable-2, White doesn’t lose stones during analysis). That would make this group alive and this string to be unexploited teire. The claim is the control definition is the same in important cases so does it differ? My understanding is the AJR-control definition would also show White has control of most of that corner. It’s about control of intersections, not control of or loss of control of an entire region of territory.
I doubt Robert’s rules make a safe reference for Japanese rules, as opposed to the official J89 text. Serious differences were found in the past, and seem to differ here as well.
By accurate scores and better komi, I take that you mean (please correct me if you meant something else) that territory scoring has finer scoring granularity, which also allows a more precise setting of komi? Thus, these two are basically the same thing. I concede that the major, common forms of area scoring rules have less granularity (i.e., games most often end with an odd scoring margin), however this issue is quite simply addressed by just introducing a button.
Of course, the setting of a specific value for komi is an independent issue the can be further refined based on better data about which choice is the most appropriate. Under any of the major rules sets, I think there is strong empirical evidence that a fair integer komi does exist. Actually, I think Japanese rules, in principle, introduces another angle of doubt as to whether there even exists an integer komi that results in jigo, given perfect play, since one would have to rule out the possibility that “no result” is not the theoretically optimal outcome.
As for the other three points (dame omission, easier counting, simpler ko), I argue that area scoring rules have the clear advantage in all three.
Dame filling is not necessary with area scoring rules (but formally required for Japanese rules)
It is a common misconception that all dame (except those that preserve sekis) should be filled, when playing with an area scoring rules set, and that Japanese rules allows players to skip filling such dame. However, ironically, the reverse is actually true! When formally following the Japanese, one should fill in all dame (except those needed to preserve sekis). Of course, many people take the informal shortcut of skipping dame filling in casual games. Under area scoring rules, it is possible to leave an even number of inconsequential dame unfilled without changing the outcome of the game.
Area scoring rules have more flexible and easier counting options
First, with Japanese rules, one has to first identify seki positions and handle those differently. Second, with area scoring rules, one has at least two different ways to count:
Use area counting, which is sometimes quite easy to do, if the board is roughly split in half. Also, in this broad class of counting methods, one technique is to only count the score for one player.
Use territory counting with pass (accounting) stones. Of course, counting the difference in the number of living stones for each player is easily approached via the proxy of counting the difference in prisoners/dead stones + how many times each player has passed. Since this is an option when using area scoring, there is basically no over-the-board counting convenience exclusive to only territory scoring.
Thus, counting for area scoring is at least as easy (or more so) than for territory scoring. Of course, the aspect of area scoring that is overwhelming easier to handle is the life and death determination phase. Counting is anyways a much simpler and purely mechanical task, which is made completely trivial in online or other settings where a computer assists.
Area scoring rules have much simpler ko rules
This is especially the case for the simplified Chinese rules (as implemented by OGS and omits “no results”) and other area scoring rules that use just a superko rule.
Japanese rules not only have the simple “basic ko rule” (forbidding two-move cycles and lifting all ko bans with a pass during alternating play), but also several other further intricacies involving ko:
A special ko rule that is applicable only during the considerations of hypothetical play for life/death determination. Specifically, in this situation hypothetical ko threats are essentially nullified by the requirement that a ko can only be retaken by passing specifically for that ko. This technicality is essential for properly handling things that are ruled dead during life/death determination, even though they cannot be captured during normal alternating play with just the usual ko rule.
Special rules to handle the potential for infinite cycles. Some cyclical cases cause the game to end in a “no result”, while other seemingly endless cycles should not cause the game to end in a “no result”, because the cycle is unbalanced (such as with sending two, returning one), or if a player is simply cycling a position that really should not be endlessly cycled (for example, repeatedly asking for game resumptions just to cycle a double-ko seki in order to delay accepting a loss).
Frankly, even explaining all of the intricacies regarding the ko rules in the Japanese rules seems beyond my understanding. Recently, as I’ve already referenced above, there was a discussion in another thread about the status of a moonshine life position. There was plenty of confusion about whether or not it should be a seki or possibly even a no result. Honestly, I have a tough time rationalizing and explaining why moonshine life should be ruled as it is under the Japanese rules and basically have to just point to the given some examples in the official commentary as authoritative and practically treat them like special rulings for special cases.
It’s one thing to say that just the basic ko works fine in most common situations, but there is still a problematic gap in the rules, if there are all these rare (but still possible) cases that require additional complexities and special rules. I think if one wants to consider the complexity of Japanese ko rules, one needs to consider these all as a whole. Otherwise, we’d be left with an incomplete or approximate description of the rules that do not correctly specify how to handle all situations.
Which definition? The original control ideas go back to (at least) 2002 and LJRG by Pauli, and the wordings here were mine, in the context of whether territory can be defined without defining alive/seki/dame in general.
I feel that you overestimate the acceptable complexity of rules. “Just” introducing a button is already at the limit (and would probably still be rejected and considered too much burden for its gain, and going against the nature of area).
Simple and well known rules are the capture rule, the ko rule, two pass stops and agreement about removals - this is the acceptable level of rule utility per complexity. Any additional rule that is not absolutely necessary will likely be seen as (and imo actually is) bloat, and unlikely to gain support in practice.
Better komi is related to granularity but not the same (for area neither 7 nor 7.5 is as fair and problemfree as 6.5 for territory).
Your other points refer to Japanese rules, while the context were territory<>area - not the same thing. Actual dame play is not required even by J89 (as we discussed here). Superko is concise but a conceptually complex addition, and unusable in an universal ruleset with dual area+territory support. And a long cycle rule under territory is just a convenient shortcut.
Ok, but I agree with Robert Jasiek that control is a concept applying to intercepts not to regions of territory (as seems to be applied by Pauli in the original). If you try to do it by regions it doesn’t make sense for un-played teire cases however.
In the game example I gave above White can easily make two eyes in the region where no territory was scored (well apart from being connected to other living areas). So if you apply the control concept to this region your choices are either to award the whole area to White for the whole area, including the two stone string. There is no choice to treat the status of these surrounding stones separately.
This is effectively acknowledged in this manner,
“2. that defects (no matter how “obvious”) have to be fixed during
the game even if neutrals are not filled (because one has to
anticipate the permanent stones)” - LJRG
But obviously in some cases players will go to scoring with unfixed defects and the rules don’t work very well if the response of the rules is to say they don’t really deal with this situation.
Edit: I also think Pauli actually defines control by intersections in the detailed description and only in the readable summary of the description uses whole regions of territory.
I fully understand that in any practical scenario, it is very unlikely to see a button be widely adopted. Maybe it is actually somewhat more likely for something like Adjusted-Lasker-Maas to become more widely adopted, since it appears to behave in a manner similar to territory scoring (and does not have a button), but (at least according to some discussion on Sensei’s Library) is strategically equivalent area scoring with a button.
Based on this sort of statement alone, it would almost seem like you would be in favor of something like the New Zealand or Tromp Taylor rules. However, I, of course, understand that is not the case. I think that our fundamental differences revolve around how we interpret something like “additional rule that is not absolutely necessary”.
A claim like this requires more justification and precision. What do you mean by “fair” and “problem free” exactly? With various common area scoring rules (like AGA, NZ, simplified Chinese, i.e., specifically those without “no results”), I believe there should exist some integer komi that results in a “fair” game, where by “fair”, I mean that the outcome under ideal perfect play will be a jigo. However, with Japanese rules, I think that with the non-integer komi of 6.5, the outcome under ideal perfect play will either be “black wins”, “white wins”, or maybe even “no result”. What are you basing your claims on?
I don’t think it makes sense to characterize superko as “conceptually complex”. Relative to the ko rules involved in the Japanese rules, it seems not only more concise, but also much easier to understand.
I think that superko is clearly usable within many practical and widely used rules sets. Nearly all of the widely used area scoring rules use form of superko rule. I’m not entirely sure what you mean by a “universal ruleset with dual area+territory support”, but I take this to mean that you are referring to some hypothetical rule set that achieves some sort of unification between area scoring and territory scoring. I’m not too sure exactly what that would even mean, and would appreciate clarification on such a concept of this “universal ruleset”, in general.
Control does apply to intersections, but scoring doesn’t want to score (unbordered) intersections individually. That would be too unconventional. Territory commonly means a surrounded region where your grasp is strong enough.
By fair komi I meant close to 50% chance in human play (not integer komi for perfect play). This is 6.5 for territory and 7 for area - but the latter has its problems while the standard 7.5 is farther from 50%.
As mentioned earlier, I think ideally the core rules of play, ko, long cycles, legal moves should be indentical between area and territory scoring, and in all game phases. And most players don’t learn any special rule inventions - they play by the above basic rules (capture, ko, removal by agreement). Thus those rules of play should also honor this minimal or “normal go” behavior as closely as possible - while at the same time offer option for both area and territory scoring.
Unlike AGA which can only score area even when counting territory, I mean true duality. The two methods would give slightly different scores, but choosing on a per-game or per-tournament basis seems fine (like done currently with J/C rules). It’s not like either scoring will disappear anyway, so proper support for both seems necessary for a modern/future ruleset (while improving J/C problems).
But all the button does is keep track of the first pass. As soon as score is kept (either on paper or electronically) a physical button (or even a screen image representing one) is not necessary. That takes care of all online games, right there. To me that is a very simple rule. Now, I understand that it applies in every game, unlike the obscure points of the Japanese rules. It’s kind of a Pascal wager: given complexities c1 and c2, and frequencies f1 and f2, in what relation do c1 * f1 and c2 * f2 stand?
And note carefully that a disputable position need not occur on the board to affect a game, just like for example in chess a stalemate almost never occurs, but affects strategy in the most profound way.
I’m not a rules expert and I basically don’t care very much for conciseness just for the sake of conciseness.
For me the most important things are that in the vast majority of games the rules are clear, unambiguous and easily applicable in practice, with minimal bookkeeping and other hassle (pass stones, superko, buttons) or obvious balance issues (7.5 komi).
Also, I prefer playing go by the tradition of territory scoring. I like its score granularity and I like its “rewarding” of optimal timing in finishing a game.
I think the job of written Japanese rules is to follow that tradition as closely as possible, while still trying to adhere to the criteria above as much as possible.
The written J89 rules sort of fail to be clear and unambiguous. I blame this on shortcomings of the writers of those rules, not on shortcomings of the principles of territory scoring.
If many of the ambiguities in written rules can be avoided by leaving out the concepts of life & death and only talk about territory, I’m all for it. Indeed life & death are more concepts that emerge from the rules, rather than actual parts of the rules.
I suppose one issue is scoring final positions where there is a bent-4 as well as unremovable ko threats. I think in such situations there is no way to reconcile traditional territory scoring with area scoring, so a choice would need to be made.
I suppose another issue that can’t really be reconciled is scoring surrounded regions in seki.
Personally, I’d be willing to break with tradition of territory scoring in such cases, if it makes a big difference in making the written rules more clear and unambiguous.
But if this would require more drastic rule changes that affect the alternating play phase more (such as requiring some variant of superko and/or pass stones), it would be too much for me. Reconciling area scoring and territory scoring is not really important to me. Using a button is just about borderline acceptable to me, but not more.
That’s a good definition when talking about life and death in the context of strategy during the game. But when talking about rules after the game, the words “alive” and “dead” are used for groups which are still on the board, meaning that the players didn’t bother capturing them during the game, so “dead” doesn’t just mean “can be captured” but really “can be captured for free”. If you can capture some stones, but you’d have to make a trade to capture these stones, then the rules don’t declare those stones as “dead” because you have to choose whether you really want to kill them (and pay the cost) or not.
There is some discussion on Sensei’s Library that asserts that the “Adjusted-Lasker-Maas” rules are actually strategically equivalent to button go (area scoring rules with a button mechanism). I haven’t verified this carefully, but it seems plausible to make this argument.
Lasker-Maas rules do already share certain key design choices with area scoring rules sets, such as adopting superko and applying the concept of playing out to settle any life-and-death disputes in the “second phase”. These choices already make the life-and-death statuses of various tricky situations behave like they would under area scoring, rather than how they would behave under the official Japanese rules. Thus, while the design of Lasker-Maas retains the scoring granularity of territory scoring, it does significantly diverge from how the common territory scoring rules (i.e., Japanese and Korean) would behave in such situations. However, I do think that it is still valid to interpret the Lasker-Maas as a form of territory scoring, but with a second (life-death resolution) phase that essentially allows “free” territory-filling plays. I’m just pointing out that I think the Lasker-Maas rules depart very significantly from the overall spirit of the Japanese rules, as I view the special handling of various situations as a core part of the Japanese rules, since I believe their entire design was meant to incorporate traditional and historic precedence.
I think the provision of playing from the prisoner piles in the second phase of Lasker-Maas is practically equivalent to using pass stones, and taking the territory count when pass stones are employed (through the entire game) essentially gives a result equivalent to the area count.
However, Lasker-Maas does not employ this pass-stone-like provision throughout the entire game, but only during this second phase. In the Adjusted-Lasker-Maas rules, the second phase is started immediately after the first pass of the game. Thus, I think you can view this first pass as being allowed “for free” without being accounted for with a pass stone, while subsequent passes are like those paid for with a pass stone (since they are not board plays that reduce the prisoner count). Essentially, I think this first pass is strategically acting like taking the button.
The above is just a hand waving sketch of what I think the equivalence argument is about, but I’m not being entirely precise and I have not carefully analyzed everything. Of course, there are still some subtle differences between pass stones and how the Lasker-Maas second phase with board plays reducing the prisoner count. For example, I think the Lasker-Maas procedure avoids the need for the provision of “white passes last” in order to prevent pass fights.
Thus, if it is indeed true that an exact strategic equivalence exists between Adjusted-Lasker-Maas rules (which are reasonable to interpret as a form of territory scoring) and button go (which is a variant of area scoring), then I think the design of these rules will have demonstrated an elegant form of unification between territory and area scoring (albeit, while abandoning the historic precedence of how Japanese rules handles various situations, such as its life-and-death determinations and the possibility of “no results”).