Life In 19x19 http://www.lifein19x19.com/ |
|
User-friendly Reading of the Japanese 1989 Rules http://www.lifein19x19.com/viewtopic.php?f=45&t=694 |
Page 1 of 10 |
Author: | RobertJasiek [ Sun May 30, 2010 5:14 am ] |
Post subject: | User-friendly Reading of the Japanese 1989 Rules |
Unlike http://home.snafu.de/jasiek/j1989c.html, this is not a detailed commentary, and unlike http://home.snafu.de/jasiek/j2003.html, this is not detailed and complete interpretation of the intention behind the rules of play. Rather the reading of the Japanese 1989 Rules below is kept simple in language and is simplifying but does not hide the essential problems. "User-friendly" is an euphemism though because Japanese rules are very difficult, J1989 are much more difficult than necessary for Japanese style rules and J1989 hide many necessary aspects behind omission in their text. *** Basic Rules - The usual fundamental rules apply. - The game consists of the phases 1. Competition, 2. Analysis, 3. Scoring. - The Competition is a sequence of alternating moves. - A move is either a play or a pass. - Two successive passes end each sequence of moves. - Removed stones become prisoners. - Suicide is prohibited. - The basic ko rule prohibits a player to make a play just after which the position just before the preceding opposing play would be recreated. Independently Alive - The only considered type of life is being independently alive. - The concept independently alive relies in two other concepts: force and two-eye-formation. - "force" is a technical term used in the Japanese 2003 Rules. It can be understood also intuitively though: That a player forces something means that he can and does choose his moves well enough to always achieve it - regardless of how the opponent replies. - A "two-eye-formation" consists of one or several strings of the same player and exactly two empty intersections so that these properties are fulfilled: 1) Each of the strings is adjacent to each of the two intersections. 2) None of the strings is adjacent to another empty intersection. 3) Each of the two intersections is adjacent only to the strings. - A string is "independently alive" if its player moving second can force to get a two-eye-formation on at least one of its intesections. Analysis - The Analysis consists of these steps: 1. Determination of the independently alive strings. 2. Removal. 3. Determination of territory. - Each string is analysed separately. It is determined whether or not it is independently alive. - For analysis of each string, imagined move-sequences starting by the attacker are considered. The defender shall try to prove "independently alive" while the attacker shall try to prove "not independently alive". Until the truth has been revealed, yet more sequences must be considered. - During the step Removal, one considers the connected regions that are adjacent only to one player's independently alive strings and that consist of intersections being empty or having opposing not independently alive strings on, which are removed. - During the step Determination of territory, a player's territory consists of the intersections of connected empty regions adjacent only to his independently alive strings. Scoring - Territory Scoring applies according to the territory determined in the Analysis. - The score is visualized by means of Japanese Fill-in Counting. Long Cycle Repetition - During Competition, a cycle ends the game exceptionally and immediately. It is, however, tolerated that the players notice occurrence of a cycle delayed. - During Competition, a cycle with an equal number of new prisoners lets the game result be a tie. For traditional reasons, this is also called "No Result" or "Neither Victory Nor Defeat" and the players are perceived to agree on the fact that such a cycle has occurred. - During Competition, a cycle with an unequal number of new prisoners lets the game result be a win for the player with the smaller number. For traditional reasons, it is tolerated though that the players perform the cycle up to ca., say, a thousand times because the player having collected enough prisoners can at some time depart, let the opponent control the entire board but still win by Scoring. - During Analysis, every sequence of moves leading to a cycle is treated as if ending upon completion of its first cycle. No stone played during such a cycle is "independently alive" though. This is equivalent to the original rules' implicit assumption of infite recurrence of that cycle. Analysis: Corrected or Simplified Aspects - The "enable" concept in the original rules leads to mistaken scores of some positions. Therefore it is not used but replaced. - A replacement for "enable" could use concepts of "capturable life" like in the Japanese 2003 Rules. This would be unnecessarily complicated though. The only advantage of this approach is to get a tool for distinguishing alive strings in "sekis" from "dead" strings. Japanese rules excempt sekis from having territory. For the purpose of applying rules and determining the winner, it is therefore superfluous to identify sekis at all. Since we do not study strategy, application of the rules can be simplified here. - An equivalence is known between always using "capturable life" versus using it only for sekis. Hence a next replacement step can be made: For "independently alive", it suffices to consider its relation to the concept "two-eye-formation". - Due to the simplifying replacement, artificial terms of or missing in the original rules like "eye-point" or "in-seki" need not be considered, either. Ignored Rules and Rules Aspects - Trivial, easily corrected flaws are ignored. - All tournament rules, komi and resignation are ignored. - All procedurally superfluous extra game ending rules are ignored. - The hypothetical ko rule is ignored. Instead the basic ko rule applies also during the Analysis. A "triple ko with one eye and one external ko" becomes a precedent: The strings of the player having the "eye" are independently alive while the opposing strings are not independently alive. The combination of a "double ko death" and a "basic ko fight about whether to dissolve it like a teire" elsewhere on the board becomes a precedent: In the double ko death, the strings of the player having the "eye" are independently alive while the opposing strings are not independently alive. The ko stone in the basic ko is not independently alive. EDIT 1: Add "independently alive" in territory. EDIT 2: Replace "permanent" by "independently alive". In the precedents, "alive" is replaced by "independently alive" and "dead" by "not independently alive". Minor corrections. |
Author: | Cassandra [ Sun May 30, 2010 6:08 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
Well done for the first edition. One suggestion: My preference would be to have "2. Removal. 3. Determination of territory." moved to Scoring. Because this is done after Analysis' real work has been finished for all chains (what I think is "1. Determination of the independently alive strings.") and only provides a combination of already found results. "1. Determination of the independently alive strings." might be complex and - in my opinion - Analysis should be kept free from such low level jobs. One question: Does "precedent" mean that the specified results cannot be found by the "usual" procedure during Analysis ? Would be an interesting choice to simplify the topic. And a valid one. |
Author: | RobertJasiek [ Sun May 30, 2010 6:23 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
Although it is just a cosmetic question where to put Removal etc., I dislike it under Scoring because previously I have used the word scoring consistently for the defining mapping from the considered position and possibly its status aspects to a score. Without the J2003-hypothetical-ko-rules (would have to check whether J1989-hypothetical-ko-rule(s) also work here), the two precedental shape classes lead to different status assessments for at least one string. Note my small edits in the initial post! |
Author: | Cassandra [ Sun May 30, 2010 6:48 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
RobertJasiek wrote: Although it is just a cosmetic question where to put Removal etc., I dislike it under Scoring because previously I have used the word scoring consistently for the defining mapping from the considered position and possibly its status aspects to a score. OK, it's just a matter of taste. Just a second suggestion: Perhaps it is more suitable to think like the user but the author of a "user-friendly" text ? ![]() Quote: Without the J2003-hypothetical-ko-rules (would have to check whether J1989-hypothetical-ko-rule(s) also work here), the two precedental shape classes lead to different status assessments for at least one string. I think I had written (years ?) before, that having precedents is a valid method to simplify the rules' text. Usually these precedents rely to positions that do not arise so often in actual games. And they give you some chances to exploit:
|
Author: | Cassandra [ Sun May 30, 2010 6:53 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
RobertJasiek wrote: Note my small edits in the initial post! Some discussions in English seem to bear more fruits than those in German. ![]() |
Author: | RobertJasiek [ Sun May 30, 2010 7:11 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
If you reveal your real name and user names elsewhere, then we have a chance to verify whether indeed you have said something years ago... Until proven otherwise, I treat each local user name with an unknown background personality as a separate person. My opinion on precedents can be found in RGG or here: http://home.snafu.de/jasiek/wagcflaw.html |
Author: | Cassandra [ Sun May 30, 2010 8:24 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
RobertJasiek wrote: I'm with you as the "scientifical" version of your rules is concerned. With this version in the back, some precedents in the "user-friendly" version may simplify the usage of the rule set for the "normal" player. |
Author: | oren [ Sun May 30, 2010 8:28 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
This is a good summary of the Japanese 1989 rules in English. http://www.cs.cmu.edu/~wjh/go/rules/Japanese.html |
Author: | RobertJasiek [ Sun May 30, 2010 10:18 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
oren, you link to the official rules text and its commentary (in English, also in both editions of the Go Player's Almanac). The original rules themselves do not explain them themselves well, contain lots of flaws as a rules text and they make mistakes in explaining professional Japanese rules. Both parts of the official commentary contain many mistakes, gaps etc. In http://home.snafu.de/jasiek/j1989c.html I give a few hints about the very low quality of the official commentary. In the meantime, I have written three generations of J1989 commentaries, of which already the first came more to the point than the original. If the original had been, what you call, "a good summary", then I would not have needed 10,000 +- 5,000 hours to solve them. The official commentary was just good enough to be deceived within 40 hours to have a first rough understanding of the rules in their pretended form on the surface. A good summary of the Japanese 1989 Rules rather is: They are too weak a model of what (professional) Japanese rules are and therefore should be replaced as soon as possible. |
Author: | oren [ Sun May 30, 2010 10:40 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
Whatever issues you have with the rules, I don't really care too much. I do think it's necessary to have a link to an English translation of the Japanese 1989 rules for people looking at this forum for information. |
Author: | Cassandra [ Sun May 30, 2010 11:03 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
Generally speaking, I can understand that Robert feels so disappointed about the quality of 1989 Nihon Kiin rules. I think, I have a more indulgent view on these rule set as Robert (I will get away with that because I'm not THE rules expert), but there are at least two aspects I want to underline. 1. The text of the 1989 Nihon Kiin rules itself leaves room for speculation. 2. Even with a well-intentioned application of these rules, it is impossible to get a consistent view on what is published as results to the examples on Life & Death. What in other words means that the original application has been inconsistent. |
Author: | RobertJasiek [ Sun May 30, 2010 3:29 pm ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
oren, there is nothing wrong with just helping people to find J1989 in the first place:) Cassandra, the consistent view on the official J1989 commentary is: - Guides towards what would have been the correct approach to writing them, like in J2003 or my J1989 commentary. - Realizing that and where the authors of the official J1989 commentary made mistakes because of partly their still insufficient theoretical background and partly because of overlooked more powerful move-sequences. |
Author: | palapiku [ Sun May 30, 2010 10:08 pm ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
I don't understand, where is seki treated? It seems like strings in seki ought to be removed by these rules since they're not independently alive. |
Author: | Cassandra [ Sun May 30, 2010 11:04 pm ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
palapiku wrote: I don't understand, where is seki treated? It seems like strings in seki ought to be removed by these rules since they're not independently alive. "not independently alive" strings of what is generally known as "Seki" will not be completely surrounded be the opponent's "independently alive" strings. So it is not allowed to take these chains off the board. |
Author: | RobertJasiek [ Mon May 31, 2010 1:20 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
Seki is not and does not need to be treated. (Not even the original J1989 defines seki.) - During the step Removal, one considers the connected regions that are adjacent only to one player's independently alive strings and that consist of intersections being empty or having opposing not independently alive strings on, which are removed. The bottom marked region does NOT consist of (only) intersections being empty or having opposing (white) not independently alive strings on. Therefore no white stones are removed from it. The upper marked region does NOT consist of (only) intersections being empty or having opposing (black) not independently alive strings on. Therefore no black stones are removed from it. |
Author: | RobertJasiek [ Mon May 31, 2010 1:53 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
Consider, e.g., how the independently alive black string partitions the board: Only from this, prisoner stones of white colour are removed. Only here Black will have territory. Note that the examples also work very similarly for the Simplified Japanese Rules, except that they allow territory in sekis. http://home.snafu.de/jasiek/sj.html |
Author: | Cassandra [ Mon May 31, 2010 6:58 am ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
Edited: 06-01: #5, #6 ------------------------------------------------- Hi, Robert, One suggestion for your rules' text, one (may be temporarily) general conclusion of mine, one question on "precedence", all further explained below.
Explanation will follow the course of 1989 Nihon Kiin rules' examples of Life & Death. Please assume unmarked bordering stones as being "independently alive". I will hide examples, which evaluation will give the same result as 1989 Nihon Kiin rules. I will hide evaluation with "my ideas", when giving the same result as "user-friendly". I hope I applied "user-friendly" correct. Example #1 (Three points without capturing) Score is three points of territory for Black. ![]() ![]() ![]() Score is zero points. ![]() Score is zero points. ![]() ![]() "Traditional" score had been "three points for White". Example #2 (Seki) Example #3 (Hanezeki) Example #4 White's chains will become prisoners. Nihon Kiin rules inconsistently see the position as Seki, due to the unclear scope of "placing another stone that cannot be captured", which cannot be board-wide. Example #5 (x-points without capturing & Seki) (edited: Black's chains are "independently alive) All of White's chains will become prisoners. Nihon Kiin rules inconsistently see the position as Seki, due to the unclear scope of "placing another stone that cannot be captured", which cannot be board-wide. White's right-hand chain will become prisoners. White's right-hand chain will become prisoners. Example #6 (Position before long life) (edited: White's single stone is "independently alive") Score is 3 points for White. Nihon Kiin rules see all of Black's chains dead. Example #7 (Bent-Four) Example #8 (Triple Ko with an eye on one side) Black's chains will become prisoners. ![]() Score is zero. Usually "cycle" during Analysis is equivalent to "not independently alive". Why not in this situation here, too ? Why torpedoing the consistent structure of "user-friendly" with this sole and only exception ? Example #9 (Direct Ko) Example #10 (Approach-move Ko) Example #11 (Moonshine Ko) Example #12 (Mannen Ko) Example #13 (Seki) Example #14 (Teire for Seki) Score is zero. It is NOT necessary for White to play at "a", as with Nihon Kiin rules. Example #15 (Bent-Four & Ko as Teire) Example #16 (Double-Ko Seki & Ko & Teire) Score is zero. Nihon Kiin rules inconsistently see all of White's chains being dead, due to the undefined "collapse of the Seki". Example #17 (Bent-Four & Ko & Double-Ko Seki) Score is zero. Nihon Kiin rules inconsistently see Black's upper left 3-stone chain alive and all of White's chains being dead, due to the undefined "collapse of the Seki. Example #18 (Bent-Four & Double-Ko Seki) Black's chain in the top left corner will become prisoners. Nihon Kiin rules inconsistently see all of Black's chain dead, due to the undefined "collapse of the Seki". Example #19 (Bent-Four & Ko & 3-point Nakade) Example #20 (Bent-Four & Ko & 4-point Nakade) Example #21 (Bent-Four & Ko & 5-point Nakade) Example #22 Example #23 (Seki missed by White during "Play") Example #24 (Two-stage Ko) Score: Both Black and White will get three points of territory, so resulting it will be zero. There is no need to play at "a" for either side nor to connect at "b" or "c" to get territory for White, as stated in Nihon Kiin rules. These rules state the result as "Seki", giving a resulting score of zero, too. Example #25 (Double-Ko Seki) |
Author: | RobertJasiek [ Mon May 31, 2010 12:25 pm ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
The major reason for using "at least one" is Chris Dams's proof, which relies on this. (Maybe one can prove it also for "all" or maybe not. ALA there is no such proof, one can rely on only "at least one".) Further reasons are that J1989 uses the "at least one" approach and that examples seem to suggest it, too. I have not seen one example from a Japanese source with a comment / variation suggesting the "all" approach. The 1-eye-triple-ko-with-external-ko precedent is needed to explain J1989. As far as my personal opinion is concerned, the precedents would be thrown away faster than you can watch:) As to your #8 question, do not ask me - ask Sakai Takeshi, major author of J1989! I have not checked whether the examples under your rules or the Nihon Kiin comments are analysed correctly you. Although I have checked carefully, I think under "user-friendly" you have made the following mistakes: #5, middle B string: independently alive! #6, single W stone: independently alive! These examples I have not checked at all for "user-friendly" so far because several move-sequences have to be studied carefully for each string in each example: #8 (without precedental rule), #16, #17, #18, #23. |
Author: | Cassandra [ Mon May 31, 2010 12:48 pm ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
RobertJasiek wrote: The major reason for using "at least one" is Chris Dams's proof, which relies on this. (Maybe one can prove it also for "all" or maybe not. ALA there is no such proof, one can rely on only "at least one".) Further reasons are that J1989 uses the "at least one" approach and that examples seem to suggest it, too. I have not seen one example from a Japanese source with a comment / variation suggesting the "all" approach. Yes, there may be some suggestion, but I think that this approach cannot be correct (at least I don't like it). And it is a source for inconsistencies within the examples. For example, "life" is made for the single White stone in example #1 through playing White stones, where proviously no White stone had been. Quote: I have not checked whether the examples under your rules or the Nihon Kiin comments are analysed correctly you. Although I have checked carefully, I think under "user-friendly" you have made the following mistakes: #5, middle B string: independently alive! #6, single W stone: independently alive! These examples I have not checked at all for "user-friendly" so far because several move-sequences have to be studied carefully for each string in each example: #8 (without precedental rule), #16, #17, #18, #23. Some examples had been a kind of snapshot, so I will check again. Just after a first look at #5: You are right with the status of Black's middle chain. But then White's 3-stone chain on the right will become prisoners. |
Author: | RobertJasiek [ Mon May 31, 2010 11:15 pm ] |
Post subject: | Re: User-friendly Reading of the Japanese 1989 Rules |
An approach with such a great success to unify two life systems should be called correct:) What do you call inconsistencies? First define what you mean, then apply it. #1: Life of the single stone: That is part of the nature of capturable-2. If you really want to be all that consistent, then go all the way to its end and define: Consistent is what already is a two-eye-formation. (And then I will ask you: Why not dissolve all sekis and disturbing lifes until we have partitioned the board into two-eye-formations?!) |
Page 1 of 10 | All times are UTC - 8 hours [ DST ] |
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |