Life In 19x19

tiger314 wrote:Funny thing is, that a 4-pass rule is actually implemented in the AGA ruleset,

Where is this documented?

I've heard this before, but I've yet to see it documented, so it not my understanding of AGA rules.

tiger314 wrote:

skydyr wrote: [...]the Japanese rules are the most arbitrary of the bunch. For example bent 4 in the corner is dead always just because. Nevermind the actual situation. Not counting eyes in seki is at odds with what pretty much everyone else does, and they have a history of situational rulings based on politics rather than anything else.

I agree with everything on that list and I would like to add the trouble of explaining hypothetical play to non-players and beginners.

The guilty secret is that hypothetical play is why bent 4 in the corner is dead. :p

tiger314 wrote:

skydyr wrote:First off, which japanese rules?[...]

I've noticed a lot of people ask which ruleset, but have there actually been any contemporary disputes caused by a player assuming a different Japanese-style ruleset was used? There seems to be next to no difference between these rulesets.

Well, if you're playing in a tournament it should be made explicit which rules, whether they are the current ones or another set. Hypothetical play, for example, is a relatively recent invention.

That said, if you are going to hold up "the japanese rules" as perfect, you really need to specify which ones, because they do have differences in how they handle edge cases. It's not like Dosaku came down the mountain with the ten Go commandments, which shall lead to a perfect contest or the wrath of God shall fall upon us. Prior to the 20th century, there were no rules, in the sense of a formal document of them, and the pass as a move did not exist. There are at least 2 codified rulesets, one from the 40s and the other from 1989, and even then when people say Japanese rules, they often mean "the rules I usually play by even though I haven't given them much serious consideration" without specifying what that actually means. To take an extreme, I could call my cat "Japanese rules" and just have disputes decided by playing from the position after she takes a few swipes at the stones on the board. There's nothing wrong with this for casual games, but if you are going to have a game where the outcome is taken very seriously, you need to be able to adjudicate in a non-arbitrary manner.

Krama wrote:

skydyr wrote:

palapiku wrote:I feel that anything beyond Japanese rules is just attempting to solve a problem that doesn't exist.

First off, which japanese rules? Second, the Japanese rules are the most arbitrary of the bunch. For example bent 4 in the corner is dead always just because. Nevermind the actual situation. Not counting eyes in seki is at odds with what pretty much everyone else does, and they have a history of situational rulings based on politics rather than anything else.

It's not just because, in chinese way of playing you can simply wait till the game is over and remove all ko threats in your area and then start the ko in the corner. Opponent has no ko threats and you kill it. Proves that bent four is dead.

But sometimes there are unremovable ko threats, like a seki one side could successfully trade.

xed_over wrote:

tiger314 wrote:Funny thing is, that a 4-pass rule is actually implemented in the AGA ruleset,

Where is this documented?

I've heard this before, but I've yet to see it documented, so it not my understanding of AGA rules.

It's in the official rules document, but it only occurs following the resumption of play after a status dispute.

Bent four in the corner shows how arbitrary traditional territory scoring rules are. Since playing the situation out would cost points (removing ko threats is likely to involve playing in territory), it always has to be evaluated locally. Under an area ruleset, you can remove ko threats free of charge (after filling dame) so the situation can be played out and unremovable ko threats, which are part of the game, can alter the status.

tiger314 wrote:Bent four in the corner shows how arbitrary traditional territory scoring rules are. Since playing the situation out would cost points (removing ko threats is likely to involve playing in territory), it always has to be evaluated locally. Under an area ruleset, you can remove ko threats free of charge (after filling dame) so the situation can be played out and unremovable ko threats, which are part of the game, can alter the status.

All rules are arbitrary. The reason to make that one is to avoid having to remove all the ko threats and get to the finish point faster.

I choose the Simple Rules
http://home.snafu.de/jasiek/simple.html
because

- rules ought to be simple and clear,
- area scoring avoids the stone scoring encore of many plays not changing the score,
- area scoring avoids the strategic simplification of stone scoring,
- for practical purposes, area scoring and territory scoring are strategically equally demanding (area scoring has the additional fight about one excess dame, territory scoring has the additional fight about avoiding typically at most one teire if the score is at most one point) but area scoring achieves this with simple and clear rules while the simplest territory scoring rules (such derived from real-world territory scoring rules related to a life concept rather than such by theorists related to pass-fights, buttons or control)
http://home.snafu.de/jasiek/sj.html
are already much less simple and clarity requires an understanding of more concepts than necessary for rules,
- the simplicity of the rules clarifies status by removals according to the regular rules of alternation instead of clarification by exceptional rules for (dis)agreements about removals (such as in New Zealand, AGA or Chinese Rules) or for life definitions (such as in the Simplified Japanese Rules),
- allowing or prohibiting suicide is equally simple and clear in rules but suicide is allowed because strategy becomes more demanding,
- passes are necessary to avoid pass fights,
- the game ends on two successive passes because this is the simplest clear game end condition and, for practical purposes, strategy is equally demanding regardless of the number of passes in a game end condition,
- passes do not lift ko bans because this gives the simplest clear game end condition and, for practical purposes, strategy is equally demanding regardless of whether passes lift ko bans,
- positional superko is used because this is the simplest clear ko ruleset and, for practical purposes, strategy is equally demanding regardless of the ko ruleset (in particular: the superko rule(s) variant).

oren wrote:

tiger314 wrote:Bent four in the corner shows how arbitrary traditional territory scoring rules are. Since playing the situation out would cost points (removing ko threats is likely to involve playing in territory), it always has to be evaluated locally. Under an area ruleset, you can remove ko threats free of charge (after filling dame) so the situation can be played out and unremovable ko threats, which are part of the game, can alter the status.

All rules are arbitrary. The reason to make that one is to avoid having to remove all the ko threats and get to the finish point faster.

All rules are arbitrary, but some rules are more arbitrary than others.
For example this position:

Why should white win? If the game continues, black clearly wins.
(Please assume the bottom left is impossible to invade)

- rules ought to be simple and clear,

I think we all agree on this one. But is it so important that there has to be no komi and all dead stones actually have to be removed by play even if players agree on the status? Plus making rules this short means a single wording error can send the game haywire.

- the simplicity of the rules clarifies status by removals according to the regular rules of alternation instead of clarification by exceptional rules for (dis)agreements about removals (such as in New Zealand, AGA or Chinese Rules)

I don't think "if there is disagreement about removeability of a group, resume play and any stones remaining on the board after next two passes are considered unremovable for counting purposes" can really be called an exceptional rule, since it pretty much says "if you don't agree what the result is, switch to simplified rules"

- passes do not lift ko bans because this gives the simplest clear game end condition and, for practical purposes, strategy is equally demanding regardless of whether passes lift ko bans,

The reason for messing about with pass to lift ko bans rules are situations like below. But, in the name of simplicity, we can probably call it black's mistake to let such a position appear and consider it rare enough to apply the simple superko.

Double button go yields the same result as by AGA rules.

Double Button Go does definitely not produce the same result as AGA rules. Taking the first button lifts the ko ban.

tiger314 wrote:

- rules ought to be simple and clear,

I think we all agree on this one.

I think that we can all agree that the rules should be clear. But simple? IMX, few games worth playing, childhood games aside, have simple rules. Furthermore, in informal play the unwritten rules are different from those in the rule book, and there are often local variants.

In go, a simple superko rule can impose a burden on human players. It is possible for the player who made a ko threat and took the superko to get lost and have to make another threat. It is also possible for a repetition to occur without the players noticing. What is the problem with rules that may be inelegant or complex but make it easy on human players?

Just to clarify

*** I get 34 pts. for White, which yields Black +14

All your scores in all positions ale using double button Go?

tiger314 wrote:Just to clarify

*** I get 34 pts. for White, which yields Black +14

All your scores in all positions ale using double button Go?

No. That was for area scoring. The 14 is a typo. It should be +13. (81 - 34) - 34 = 13.

Bill Spight wrote:

tiger314 wrote:

- rules ought to be simple and clear,

I think we all agree on this one.

I think that we can all agree that the rules should be clear. But simple? IMX, few games worth playing, childhood games aside, have simple rules. Furthermore, in informal play the unwritten rules are different from those in the rule book, and there are often local variants.

In go, a simple superko rule can impose a burden on human players. It is possible for the player who made a ko threat and took the superko to get lost and have to make another threat. It is also possible for a repetition to occur without the players noticing. What is the problem with rules that may be inelegant or complex but make it easy on human players?

The problem is that complex usually means hard on human players. I think the best example being the Ing ko rule. The ko section of Ing's rules has like 300 words (Robert's complete rules have only about 150) and is understood by only a handful of amateur players, and I have never seen anyone successfully implement it in a program. Why isn't the less than twenty words of not repeating previous positions enough? I know superko is quite tricky to apply, but with the exception of voiding/drawing a game with a complex ko, there is nothing simpler.

tiger314 wrote:

- rules ought to be simple and clear,

I think we all agree on this one.

Bill Spight wrote: I think that we can all agree that the rules should be clear. But simple? IMX, few games worth playing, childhood games aside, have simple rules. Furthermore, in informal play the unwritten rules are different from those in the rule book, and there are often local variants.

In go, a simple superko rule can impose a burden on human players. It is possible for the player who made a ko threat and took the superko to get lost and have to make another threat. It is also possible for a repetition to occur without the players noticing. What is the problem with rules that may be inelegant or complex but make it easy on human players?

tiger314 wrote: The problem is that complex usually means hard on human players. I think the best example being the Ing ko rule. The ko section of Ing's rules has like 300 words (Robert's complete rules have only about 150) and is understood by only a handful of amateur players, and I have never seen anyone successfully implement it in a program. Why isn't the less than twenty words of not repeating previous positions enough? I know superko is quite tricky to apply, but with the exception of voiding/drawing a game with a complex ko, there is nothing simpler.

In 1977 or so I wrote a short article for the AGA Journal about the Ing rules at the time. Taiwan had adopted them. They were the same as what were later the Taylor-Trump rules, or almost so, and included a simple superko rule. In the article I recommended the use of pass stones, which I called bookkeeping stones. From what I have heard, around 1981 someone pointed out to Ing that a superko consisting of two identical double ko death positions meant that one of the "dead" groups had to be taken to leave only one double ko death, or one "dead" group could live. Apparently this was an unintended consequence for Ing. The superko rule had changed the game more than Ing had meant to do, making it more complex. (It is also likely that some pro players shared Ing's dismay and talked to him about that.) In terms that Ing later used, the superko rule turned two disturbing kos into one fighting ko. Ing revised his rules a number of times, and, as far as kos are concerned, they became more complex in order to make the game less complex. (IMO, he succeeded with the 1996 version of his rules, but they are not all that clear, because he attempted to derive his rules from general principles. Others have explained them more clearly. I imagine that the Ing rules could be programmed if you ignore the principles. ) I do not like the Ing rules, but I agree in general and within limits with the idea of making the rules more complex or "illogical" in order to make them less complex for humans.