Dispute Breaker for Japanese Rules
robert pauli
In the thread "October KGS Tournament" RJ seems to
make a fuss, OK, but let's face it, he has a point.
Any server supporting Japanense Rules - very well! BTW -
IMHO should have a dispute breaker implemented to enable
players to get to their result on their _own_, without
"help" of a third person.
I outlined an (IMHO, pretty perfect) one in the thread
"Conclusions about Japanese Rules " on 2004-09-22:
http://groups.google.de/groups?q=best+rp+bubble&hl=de&lr=&ie=UTF-8&group=rec.games.go&selm=4854c6b5.0409220225.4b3c4fae%40posting.google.com&rnum=1
With it we only have to get used doing teire unforced
(don't they belong to the skill part anyway ?) and
clicking on dead and neutral areas after game "end"
- that's all. Dame can (and should) remain unfilled.
Applied to RJ's example (shrinked again)
-----------
10 | . X . X . |
9 | . X X X X |
8 | X O O O O |
7 | X O . O . |
6 | X O O O O |
5 | X X X X O |
4 | X . O . O |
3 | X O . O O |
2 | X . O X O |
1 | X X X X O |
-----------
a b c d e
Black connects at a9,
White passes (c3 isn't really teire),
Black passes too (all moves made),
White passes (no dead or neutral black area),
Black clicks c3 (neutral white area),
White passes (no more dead or neutral black area), and
Black passes (for a result of 3 - 2).
Note that dame at b2, b4, and d4 neither
had to be filled nor tagged.
But what if White disagrees?
...
Black clicks c3 (neutral white area),
White clicks the disagree button ("No, c3 is mine!"),
White clicks c7 and e7 (helping bubbles), and
White clicks the test button - setting up
-----------
10 | * * * * * |
9 | * * * * * |
8 | * O O O O |
7 | * O . O . |
6 | * O O O O |
5 | * * * * O |
4 | * * O * O |
3 | * O . O O |
2 | * * O * O |
1 | * * * * O |
-----------
a b c d e
each * an uncapturable black stone and Black to start:
Black captures at c3,
the system recognizes a black "two-eye formation"
overlapping White's claim (part beyond *), and
Black won dispute _and_ game.
Now you know why Black hurried to fill a9:
White would have declared a9 and a10 neutral
(with one click), and Black would have to be
satisfied with jigo (since disgreeing loses).
RJ's example with the minimal endgame ko neither is
a problem: Leave the ko open - your opponent will
tag it as neutral, without fearing any dispute.
Best to all
RP
Bill Spight
Very nice. :-)
Minor nit:
> -----------
> 10 | . X . X . |
> 9 | . X X X X |
> 8 | X O O O O |
> 7 | X O . O . |
> 6 | X O O O O |
> 5 | X X X X O |
> 4 | X . O . O |
> 3 | X O . O O |
> 2 | X . O X O |
> 1 | X X X X O |
> -----------
> a b c d e
>
> Black connects at a9,
> White passes (c3 isn't really teire),
Technically, it is teire.
-----------
| . . . . . |
| b X X X X |
| X O O O O |
| X O . O . |
| X X O O O |
| . X X X O |
| a O X O O |
| O . O O O |
| . . O . O |
-----------
Both 'a' and 'b' are teire.
Best,
Bill
robert pauli
message news:<415ECE73...@pacXbell.net>...
>>
-----------
10 | . X . X . |
9 | . X X X X |
8 | X O O O O |
7 | X O . O . |
6 | X O O O O |
5 | X X X X O |
4 | X . O . O |
3 | X O . O O |
2 | X . O X O |
1 | X X X X O |
-----------
a b c d e
<<
> Technically, it is teire.
Dear Bill,
your position (below)
-----------
| . . . . . |
| . X X X X |
| X O O O O |
| X O . O . |
| X X O O O |
| . X X X O |
3 | . O X O O |
2 | O . O O O |
1 | . . O . O |
-----------
a b c d e
still is hot if there are ko threats for Black
somewhere else, so let me assume nothing else.
Now, c3 in the first and a3 in the second diagram
both indeed are teire in the sense that White can
be "forced" to play them.
But the fine difference is that
- c3 defends no territory, whereas
- a3 _does_
My scheme reveals that. With it White can (and
should) omit c3 because she won't claim it anyway
(and we want to come to an end), but White should
connect at a3 since she wants to claim that
"bubble" worth three points later.
Shouldn't these "territory-protecting" teire
(tautology? ;-) naturally be part of the game?
This would also pave the way to count games
without messing up the final position.
Best
RP
Brian Sundermeier
> Any server supporting Japanense Rules - very well! BTW -
> IMHO should have a dispute breaker implemented to enable
> players to get to their result on their _own_, without
> "help" of a third person.
>
> I outlined an (IMHO, pretty perfect) one in the thread
> "Conclusions about Japanese Rules " on 2004-09-22:
> http://groups.google.de/groups?q=best+rp+bubble&hl=de&lr=&ie=UTF-8&group=rec.games.go&selm=4854c6b5.0409220225.4b3c4fae%40posting.google.com&rnum=1
>
Brilliant. I want to make sure that I understand the implications of
this (and other) dispute breakers. From my understanding of your
recommended procedure, Robert, "three points without capturing" (Life
and Death Example 1 JRG89) would be a seki because black cannot create
a two-eyed formation completely on his bordered set and also because
white cannot create a two-eyed formation on her bordered set. Is this
correct?
It also seems that KGS has plans to implement a dispute breaker
procedure (http://kgs.kiseido.com/en_US/help/dispute.html). BTW, Does
anyone know if this is still being planned, and if so, when it will
take effect? Based on my understanding of the KGS implementation and
assuming reasonable play in the dispute phase, the white stone would
be marked as dead and the black stones marked as alive, giving black
two points of territory and one prisoner (3 points).
In contrast, with Robert Jasiek's New Amateur-Japanese Rules, "three
points without capturing" is 3 points in favor of white.
And if this position is played out according to "play-it-out rules",
it seems that the result would be 1 point in favor of White (if white
plays first).
Whew...who knew that settling disputes would be as complicated as
go?...hehe.
Brian
Robert Jasiek
wrote:
>http://kgs.kiseido.com/en_US/help/dispute.html
First a note on the Chinese dispute handling: to avoid the IGS escaper
problems, resumptions should be restricted in number, i.e. there has to
be a last resumption. IMO, it is sufficient if a game has at most one
resumption.
Besides it must be possible to resume if the opponent rejects to start
it by not also pressing the Undo button. IOW, if a player presses Undo
during the dead stones phase, then this must be enough for resumed
alternation to start.
***
Now let me comment on the plans for KGS's Japanese dispute handling.
It is a terrible design to allow "escapers" to cheat. And it is entirely
superfluous. Instead of using the players' marks of dead stones, if a
game is resumed for the hypothetical play, all marks of dead stones
should be ignored and the players should simply continue to play and
remove as many opposing stones as possible. Dead is what can be removed
during the hypothetical play.
With KGS's plan or with my modification mentioned before, KGS-Japanese
Rules with Dispute Handling would NOT be the same as (official) Japanese
rules, even if this is claimed on the KGS webpage about the planned
dispute handling. However, differences ARE scarce. E.g., a position
might get a different result if it has a bent-4 and certain kodate
elsewhere. (Differences are not as scarce as under the New
Amateur-Japanese Rules. Also if there is only one (and not two)
hypothetical play sequence, then there is no general colour-symmetry of
the final-position. One can tolerate these issues, but one should not
deny their existence.)
The suggested hypothetical ko rule together with its context of how
hypothetical play shall work is a hybrid of the J2003 hypothetical ko
rule and the New Amateur-Japanese Rules, if the latter had only one
analysis. This means that in arcane positions the results are not
studied yet and would sometimes be unexpected.
Then there is the possibility of infinite sequences in hypothetical
play. This is not treated yet.
Why not use positional superko during the hypothetical play (positions
of the alternating-sequence are forgotten)? If it is ideology not to use
superko, then why not use the Basic-Fixed Ko Rules like under the New
Amateur-Japanese Rules? Their effect has been tested reasonably
carefully by me for two analyses and also (less exhaustively) for only
one analysis sequence.
I guess that using the planned J2003 hypothetical ko rule (but beware
that J2003 has more restrictions specified in the definition of ko-pass)
for exactly one analysis can work in principle. However, it is necessary
to test that before using it on KGS. Firstly, the effect depends on the
exact wording of that hypothetical ko rule. Secondly, the numbers and
effects of passes / ko-passes plays a great role! I have not tested such
a type of hypothetical ko rule together with hybrid-passes, i.e. when
ko-passes also have an ending function. Either the ko and pass rules for
hypothetical play are worded very carefully and then tested exhaustively
or the hypothetical ko rule(s) should be such that have already been
tested (partially): Positional Superko or Basic-Fixed Ko.
***
It is great that KGS plans dispute handling. However, if it can be made
correctly at once, then there is no need to introduce new forms of
disputes by being careless. The dispute handling should be reflected and
tested carefully before it is implemented.
--
robert jasiek
Chris Dams
jas...@snafu.de (Robert Jasiek) writes:
>With KGS's plan or with my modification mentioned before, KGS-Japanese
>Rules with Dispute Handling would NOT be the same as (official) Japanese
>rules, even if this is claimed on the KGS webpage about the planned
>dispute handling. However, differences ARE scarce. E.g., a position
>might get a different result if it has a bent-4 and certain kodate
>elsewhere.
However, there could be both a "pass" and a "pass for ko" button during
hypothetical play. Ko-captures that are currently forbidden could be
indicated just the same way as during the alternating sequence: by marking
the stone that cannot be captured.
>Why not use positional superko during the hypothetical play (positions
>of the alternating-sequence are forgotten)?
Why not stay closer to J2003? OC, what is most important is that the exact
rules are obtainable somewhere (on a webpage or a server help file) but
the proliferation of many different rule sets is not good either, I think.
If the position repeats during hypothetical play, one could stop the
hypothetical sequence and treat it as if an infinite sequence was played.
>I guess that using the planned J2003 hypothetical ko rule (but beware
>that J2003 has more restrictions specified in the definition of ko-pass)
>for exactly one analysis can work in principle. However, it is necessary
>to test that before using it on KGS. Firstly, the effect depends on the
>exact wording of that hypothetical ko rule. Secondly, the numbers and
>effects of passes / ko-passes plays a great role! I have not tested such
>a type of hypothetical ko rule together with hybrid-passes, i.e. when
>ko-passes also have an ending function. Either the ko and pass rules for
>hypothetical play are worded very carefully and then tested exhaustively
>or the hypothetical ko rule(s) should be such that have already been
>tested (partially): Positional Superko or Basic-Fixed Ko.
Why not advertise the use of your own J2003 rules? These at least seem to
have been tested (by you). Well, of course, only uncapturable and capturable-1
should be implemented (Note that a stone that is alive because of snap-back
occurs regularly, at leat in my games). I guess capturable-2 is the point
where things are getting too arcane. Also, most people would not like it
when they had to fill all dame. IMO it is not so bad if the score is only
guaranteed to be the same as under J2003 if all dame are filled. One can
always do that and it only costs a few seconds in most cases. It will cost
the opponent about the same amount of time, even if (s)he sees no need to
fill all dame, because of having to hit the pass button repeatedly. If you
know that you are going to fill all dame, you should take that into account
when using your thinking time during the game. The thinking time that is
involved here will probably be less than the time margin that has to be kept
anyway in order to make sure that you have enough time to defend against
some bogus last-minute invasion by the opponent. I for myself fill dame
to the point where I start trusting the IGS (which is where I play)
scoring algorithm. As far as I know, this algorithm counts points in a
seki, but I only encountered one game in the approx. 1000 I played there,
where this would have made a difference in the score (not in the result).
>It is great that KGS plans dispute handling.
Well, as far as I know that page has been there for quite some time now,
so I am not sure how serious the plans for implementing what is described
are. On the other hand, perhaps I do not care that much, since I am not going
to play on KGS as long as there is any other go server that can be accessed
by free software, while KGS cannot.
Best,
Chris
robert pauli
message news:<1d278a37.04100...@posting.google.com>...
Dear Brain,
> From my understanding of your recommended procedure,
> Robert, "three points without capturing"
> (Life and Death Example 1 JRG89)
> would be a seki because black cannot create a
> two-eyed formation completely on his bordered
> set and also because white cannot create a
> two-eyed formation on her bordered set.
> Is this correct?
Yes, six points controled by nobody.
Full-board example, final position:
-------------
5 | . . . . X . |
4 | X X X X X O |
3 | X O O O O O |
2 | O X X O . O |
1 | . X X O O . |
-------------
a b c d e f
White won't tag the black stones, say by a click on
b1, as dead because she can't cover "her" bubble
(inside = a1,b1,b2,c1,c2) by a two-eye formation.
For the same reason Black won't tag a2 as dead -
he can't cover "his" bubble (inside = a1,a2).
Nobody will now need to tag a1 or f5 as neutral
because each "sees" both colors.
So, the score is 4 - 2 = B+2 - without filling dame
and, in this case, without any tagging!
OC, White should have captured, gaining two captives
net for a tie: a1::/b2 b1/c1:. c2/a1 b1./a2 c1
> It also seems that KGS has plans to implement
> a dispute breaker procedure
> (http://kgs.kiseido.com/en_US/help/dispute.html).
I welcome any. Thanks for the link.
> BTW, Does anyone know if this is still being planned,
> and if so, when it will take effect?
Would like to know myself.
> Based on my understanding of the KGS implementation...
We shouldn't care about their implementation, just about
their _definition_. However, from what I read, it's not
clear, for instance:
- Are there several or just one dispute possible ?
- Is the dispute about one single stone, about
a connected group of them, about several such
groups of one player, or even about several
such groups of both players ?
- How exactly are the disputed stones picked ?
- Does the test start with no ko bans,
even if one was left over from the game ?
- Three passes in a row end the test,
great - but, heck, WHO starts a test ??
- Does a post-end pass release all banned kos,
or just one ?
> Based on my understanding of the KGS implementation
> and assuming reasonable play in the dispute phase,
> the white stone would be marked as dead and the black
> stones marked as alive, giving black two points of
> territory and one prisoner (3 points).
Let me first try to understand the KGS scheme
myself, especially this part:
>> if any of the stones [in dispute] are still alive,
>> then all of the stones that were in dispute will
>> be kept alive
- "Alive" seems to mean "on the board".
- Defender can't be the one to choose the stones in
dispute - otherwise he could prove about everything
to be alive if he at least has a single defendable
stone (using above's quote), therefore:
- Attacker chooses target, and for his benefit:
- Attacker always chooses a single stone.
- Several tests are possible.
- Regarding snap back, Defender starts tests.
- Let's assume tests always to start with no ko ban
- and each pass clearing all recent ones.
With this, _all_ five stones would be dead, the four
black too, but you don't think so, hmm... OK, I see,
to meet I'd have to modify my "understanding" to
- Attacker chooses one connected group of stones.
- This whole group is alive if any stone in its
color remains on its area at test end.
This also makes sense to above's quote.
Now I agree with you: B+5 under KGS.
> In contrast, with Robert Jasiek's
> New Amateur-Japanese Rules,
> "three points without capturing"
> is 3 points in favor of white.
Faintly remember he continued from the "final"
position twice, each color starting once.
This may define a result, but I still don't
get how this categorizes stones and space in
the "final" position. We have to ask RJ.
> And if this position is played out according to
> "play-it-out rules", it seems that the result would
> be 1 point in favor of White (if white plays first).
No, White gains two - with pass stones
(and an even move tally) even three.
> Whew...who knew that settling disputes
> would be as complicated as go?...hehe.
Indeed, this rule stuff is a nightmare:
the "end" isn't necessarily the end :-)
However, those that don't care, and I don't mind'em,
will end as usual (well, almost), and those few that
care, will end just the same - except that they know
what they're skipping.
Best
RP
PS
Note that under my scheme Black can't
claim any points here
---------------------------
| X X X . . X . X . . X X X |
| O O O O O X X X O O O O O |
---------------------------
(he should have taken one side, letting the other
make life), but under the KGS scheme, as far I see,
the white stones are dead.
Therefore KGS's claim below is wrong
>> These rules are not the exact Japanese rules,
>> but they do have the same end result.
Chris Lawrence
> Note that under my scheme Black can't
> claim any points here
>
> ---------------------------
> | X X X . . X . X . . X X X |
> | O O O O O X X X O O O O O |
> ---------------------------
>
> (he should have taken one side, letting the other
> make life), but under the KGS scheme, as far I see,
> the white stones are dead.
How does white make life on one side? White looks dead to me.
--
Chris
Robert Jasiek
wrote:
>> New Amateur-Japanese Rules,
>position twice, each color starting once.
>This may define a result, but I still don't
>get how this categorizes stones and space in
>the "final" position. We have to ask RJ.
What are your questions?
--
robert jasiek
Robert Jasiek
Dams) wrote:
>However, there could be both a "pass" and a "pass for ko" button during
>hypothetical play.
Sure, however, do you expect many server players to know the differences
between "pass" and "kopass"? IMO, for server rules it is too complicated
to have three move types.
>Why not stay closer to J2003?
IMO, server players tend to be impatient. Therefore dispute rules must
be as simple as possible rather than as complicated as possible. With
J2003 there are dozens of dispute-analyses, one for each string. It is
much simpler to have only one or two dispute-analyses.
>OC, what is most important is that the exact
>rules are obtainable somewhere (on a webpage or a server help file) but
>the proliferation of many different rule sets is not good either, I think.
Right. Therefore there should be the Japanese Rules for Go Servers. I
suggest: New Amateur-Japanese Rules with the following modifications:
- one analysis instead of two analyses
- dame do not score
>If the position repeats during hypothetical play, one could stop the
>hypothetical sequence and treat it as if an infinite sequence was played.
I am not sure now if this is without side-effects.
>Why not advertise the use of your own J2003 rules?
J2003 clarify professional Japanese rules but both are not suitable for
usage in practical play: There are too many hypothetical-strategies and
-sequences. Not to mention the trivial constant factor that each string
needs its extra analysis.
J2003 can be advertised for various reasons but not for applicability.
>These at least seem to have been tested (by you).
They have. Also the New Amateur-Japanese Rules have been tested. NAJ
with only one analysis have also been tested by me, but much less
carefully.
>Well, of course, only uncapturable and capturable-1
>should be implemented
Capturable-2 is too important for being not implemented.
>Also, most people would not like it when they had to fill all dame.
This seems to be the major point of Japanese-style rules on go servers.
It does not require the Logical Japanese Rules of Go with their
side-effects that are much more frequent than those of NAJ with the
above modifications.
>IMO it is not so bad if the score is only
>guaranteed to be the same as under J2003 if all dame are filled. One can
>always do that and it only costs a few seconds in most cases.
Under some rulesets, the problem is to have a dame filling phase in
between alternation and hypothetical play. Otherwise filling of dame
would have to be done already during the alternation, if a player wants
to avoid side-effects.
--
robert jasiek
Erik van der Werf
White claims life ==> Black should refute the claim by capturing. When
Black choses to attack one side the other side makes two eyes. So maybe
we should say that each white group is 50% alive?
E.
PS How would you score the simplified position below?
---------------------------
| . . . O X . X . X O . . . |
| O O O O X X X X X O O O O |
---------------------------
Chris Dams
jas...@snafu.de (Robert Jasiek) writes:
>IMO, server players tend to be impatient. Therefore dispute rules must
>be as simple as possible rather than as complicated as possible. With
>J2003 there are dozens of dispute-analyses, one for each string. It is
>much simpler to have only one or two dispute-analyses.
OC, one would only analyze strings for which there is a disagreement on
their status. On the other hand, you certainly have a point. Cheaters may
start disputing the status of each and every string....
This could easily take an hour, I suppose ... :-(
>>If the position repeats during hypothetical play, one could stop the
>>hypothetical sequence and treat it as if an infinite sequence was played.
>I am not sure now if this is without side-effects.
I would think that it is provable that this is without side effects.
Consider every hypothetical-sequence that is compatible with a particular
hypothetical-strategy. Then map finite hypothetical sequences that repeat
into ones with repetition removed by removing the part between the
first reoccurance of a situation (a situation consists of position, player to
move and set of intersections that are currently forbidden for a player by
hypothetical ko-rule). Infinite hypothetical sequences have a situation
that occurs over and over. Then we modify infinite hypothetical sequences
by taking the move sequence between first and second occurence of the
first reoccuring situation and use it also between second and third, third
and fourth occurence and so on. Also, if there is repetition between these
reoccuring situations, this is removed in the same way as for finite
sequences. I would think that this way the set of all left parts of
hypothetical sequences is mapped into a new hypothetical-strategy that
forces exactly the same as the original one and furthermore also forces
the same if we add the proposed server rule that after first repetition of a
situation in the hypothetical sequence the sequence is treated as infinite.
This would prove that the proposal is without side-effects, but I do not
feel very much like working out every episilon and delta in this proof. I
am a physicist, you know. Physicists just make things plausible. ;-)
>>Well, of course, only uncapturable and capturable-1
>>should be implemented
>Capturable-2 is too important for being not implemented.
Huh? Above you said that there should not be "pass for ko" and now
capturable-2 should be implemented? I would think situations in which
"pass for ko" can make a difference occur far more often than situations
where capturable-2,3 makes a difference.
>>Also, most people would not like it when they had to fill all dame.
>This seems to be the major point of Japanese-style rules on go servers.
>It does not require the Logical Japanese Rules of Go with their
>side-effects that are much more frequent than those of NAJ with the
>above modifications.
Yes. I think this is a good argument in favour of your proposal.
Best,
Chris
Chris Lawrence
> > > Note that under my scheme Black can't claim any points here
> > > ---------------------------
> > > | X X X . . X . X . . X X X |
> > > | O O O O O X X X O O O O O |
> > > ---------------------------
> > >
> > > (he should have taken one side, letting the other make life), but under
> > > the KGS scheme, as far I see, the white stones are dead.
> >
> >
> > How does white make life on one side? White looks dead to me.
>
> White claims life ==> Black should refute the claim by capturing. When Black
> choses to attack one side the other side makes two eyes. So maybe we should
> say that each white group is 50% alive?
I see now, black can't kill both groups or, conversely, white can't save
both groups. Isn't that just like any other position on the Go board as
the game develops? You often see a kind of miai for the strength of two
moves ocurring - black can either make the enclosure or play an
extension but not both, white can prevent one but not both. How is this
conceptually any different to the end of the game? How is this any
different to asking CGoban2 to 'score' any other balanced situation? In
the limit this thinking seems to lead to CGoban2 playing a hypothetical
perfect game from an emtpy board and scoring the result while the
players watch.
I don't see why you would say that each group is 50% alive. Surely one
group is alive and the other dead. You might as well say that in a game
every stone is 34% alive, on average, for a given player. In fact that
might an interesting way to gauge some aspect of the 'goodness' of a
player (Mr X has a 92% aliveness of his stones compared with 72% last
year).
> PS How would you score the simplified position below?
>
> ---------------------------
> | . . . O X . X . X O . . . |
> | O O O O X X X X X O O O O |
> ---------------------------
Black is alive, white has one dead and one live group, assuming both
sides are playing Go and trying to win (a reasonable assumption I should
think). Black has 15, white has 2. Of course the point of no return I
mentioned previously has not passed, so CGoban2 should have no way to
guess the situation unless it has some basic reading and assumptions
built in.
There's nothing to stop black killing both groups while white continues
to pass. There's nothing to stop white making two living groups while
black continues to pass. In practice the players will score the groups,
or, far more likely, play the additional two moves which take it past
the point of no return and remove the uncertainty. Having watched this
thread with interest I am completely failing to see any problem with
CGoban2 or the way people are playing on KGS. Robert seems to be trying
to find ways to weasle points out of rules disputes, something which I'm
sure is against the whole spirit of the game and never the intention of
any rules creators.
--
Chris
Ted S.
wrote in news:Pine.BSF.4.60.04...@newred.gradwell.net:
>> PS How would you score the simplified position below?
>>
>> ---------------------------
>> | . . . O X . X . X O . . . |
>> | O O O O X X X X X O O O O |
>> ---------------------------
>
> Black is alive, white has one dead and one live group, assuming both
> sides are playing Go and trying to win (a reasonable assumption I
> should think).
Are you sure that's a reasonable assumption on rec.games.go? :-)
--
Ted <fedya at bestweb dot net>
Barney: Hey, Homer, you're late for English.
Homer: Who needs English? I'm never going to England.
<http://www.snpp.com/episodes/7F12.html>
Robert Jasiek
<new...@holosys.co.uk.invalid> wrote:
>I am completely failing to see any problem with CGoban2
CGoban / KGS using KGS-Japanese Rules have in particular the following
problems:
- They do not inform the players about the rules properly. E.g., the
players should be informed about the meaning of marks / missing marks on
intersections.
- The rules do not agree to official Japanese rules while they
information about them pretends that they would agree.
- Teire behave unpredictably. This is not documented.
- The special nature of basic endgame kos is not documented.
>or the way people are playing on KGS.
See earlier messages in this and other related threads.
>Robert seems to be trying
>to find ways to weasle points out of rules disputes,
I am trying to minimize the number of necessary disputes by getting as
much reliable information about the KGS-Japanese Rules as possible. To
recall, the KGS-Japanese Rules state explicitly: "If players cannot
agree whether a group is alive or dead, they must accept the judgement
of a third neutral player." This means that a disagreement about life
and death leads to a dispute. Disputes are a common part of the rules.
This is especially so because the most difficult part of applying
Japanese style rules is determination of life and death and because the
rules do not state anything else about life and death. There are only
two options: either the players agree or the players have a dispute.
When you say the above, do you mean that I ought to agree to my
opponent's assessment of life and death always and unconditionally?
>something which I'm
>sure is against the whole spirit of the game
What is "the whole spirit of the game" IYO? Is this "always and
unconditionally agreeing to one's opponent's assessment of life and
death"?
>and never the intention of any rules creators.
Rules creators that allow either agreement or disputes have wanted to
allow either agreement or disputes, unless they have been deceiving
everybody intentionally or have been extraordinarily careless.
***
You sound like wanting different rules as follows: "Alternate. Then
pass. Then either agree or resign before your opponent gets a chance to
resign."
--
robert jasiek
Chris Lawrence
> On Tue, 5 Oct 2004 16:24:36 +0100, Chris Lawrence
> <new...@holosys.co.uk.invalid> wrote:
> >I am completely failing to see any problem with CGoban2
>
> CGoban / KGS using KGS-Japanese Rules have in particular the following
> problems:
> - They do not inform the players about the rules properly. E.g., the
> players should be informed about the meaning of marks / missing marks on
> intersections.
I can't think of any game at all at any level which I've watched on KGS
in 18 months which caused any problem. The 'missing' marks seem
obvious, and the players make a judgement and agree on the result.
> - The rules do not agree to official Japanese rules while they
> information about them pretends that they would agree.
I doubt the vast majority could care less. You seem to think that rules
are everything, in every aspect of your psyche as far as I can see.
From proposing defined tags in Subject headers to arguing over the exact
score of a half point ko in a free game with more than 100 points
difference. It's like you're really insecure without the security
blanket of rules for everything and that you don't so much play Go but
play rules with your opponents, leaving them bemused and frustrated at
times it seems.
> - Teire behave unpredictably. This is not documented.
> - The special nature of basic endgame kos is not documented.
I don't know, Go has always felt very simple to me, even simpler than it
did when I first encountered it, even though it's a different game to me
now. It always feels to me like everyone I meet sees a painting while
you see discrete regions of individual wavelengths of light from
specific co-ordinates within the picture frame.
> >or the way people are playing on KGS.
>
> See earlier messages in this and other related threads.
See my earlier replies in response.
> >Robert seems to be trying
> >to find ways to weasle points out of rules disputes,
>
> I am trying to minimize the number of necessary disputes by getting as
> much reliable information about the KGS-Japanese Rules as possible.
I really don't think that most people could care about the minutae of
the rules and their sub clauses. Everyone I've met mostly play to the
Japanese rules on KGS without any problem.
> To recall, the KGS-Japanese Rules state explicitly: "If players cannot
> agree whether a group is alive or dead, they must accept the judgement
> of a third neutral player."
Sound sensible to me. If you can't work out the status then get a
stronger player to help. What's wrong with that? Seems to work well
for all these pros I hear about.
> This means that a disagreement about life and death leads to a
> dispute. Disputes are a common part of the rules.
Especially if you engineer such situations :(
> This is especially so because the most difficult part of applying
> Japanese style rules is determination of life and death and because
> the rules do not state anything else about life and death. There are
> only two options: either the players agree or the players have a
> dispute.
>
> When you say the above, do you mean that I ought to agree to my
> opponent's assessment of life and death always and unconditionally?
No, it means that you're like a obsessive legal pedant who nitpicks
endlessley over small things which most people just get on with.
> >something which I'm
> >sure is against the whole spirit of the game
>
> What is "the whole spirit of the game" IYO? Is this "always and
> unconditionally agreeing to one's opponent's assessment of life and
> death"?
Why do you keep asking me questions and preloading the answer in your
question? It says a lot about your thought processes. The spirit of
the game... having fun, respecting your opponent, respecting your
environment, communicating with people, not having an underhand agenda,
etc.
> >and never the intention of any rules creators.
>
> Rules creators that allow either agreement or disputes have wanted to
> allow either agreement or disputes, unless they have been deceiving
> everybody intentionally or have been extraordinarily careless.
You can either see the painting or see the discreet patches of varying
wavelengths. The former is more than the sum of the latter.
> You sound like wanting different rules as follows: "Alternate. Then
> pass. Then either agree or resign before your opponent gets a chance to
> resign."
Stop assuming that I want this rule or that rule. I couldn't care less
about all these rules and clauses an sub-clauses. There's so much stuff
for me to learn for the next several years without needing to ever look
at a single rule. I wish you could see that.
--
Chris
Chris Dams
> ... I would think that this way the set of all left parts of
>hypothetical sequences is mapped into a new hypothetical-strategy that
>forces exactly the same as the original one ...
This is not true, of course. It is easy to think of a position in which
a player can force either capture a particular string or a non-periodic
infinite hypothetical sequence. However, by my procedure a non-periodic
infinite hypothetical sequence would get mapped into a periodic one. Hence
under ordinary J2003 a player could force either capture of some stones or a
non-periodic infinite hypothetical sequence, while under my modification
this is not possible. On the other hand, why would anyone want to force
that...? The things in which J2003 is interested in forcing seem to be
better behaved. Still, it appears that one has to be carefull...
Robert Jasiek
Dams) wrote:
>This would prove that the proposal is without side-effects
Which you have not defined so far;)
>I would think situations in which
>"pass for ko" can make a difference occur far more often than situations
>where capturable-2,3 makes a difference.
Do you have "frequent" examples (more often than 1:800) where using a
ko-pass rule makes a difference compared to using the basic-ko rule?
--
robert jasiek
Robert Jasiek
<new...@holosys.co.uk.invalid> wrote:
>You seem to think that rules
>are everything, in every aspect of your psyche as far as I can see.
When lacking factual arguments, change topic by discussing psyche and
moral
>it means that you're like a obsessive legal pedant who nitpicks
>endlessley over small things
and by being "polite"...
--
robert jasiek
Chris Dams
jas...@snafu.de (Robert Jasiek) writes:
>On Tue, 5 Oct 2004 14:36:50 +0000 (UTC), chr...@gamow.sci.kun.nl (Chris
>Dams) wrote:
>>This would prove that the proposal is without side-effects
>Which you have not defined so far;)
"Without side-effects" means that in every final-position the same
groups would be dead or alive.
>>I would think situations in which
>>"pass for ko" can make a difference occur far more often than situations
>>where capturable-2,3 makes a difference.
>Do you have "frequent" examples (more often than 1:800) where using a
>ko-pass rule makes a difference compared to using the basic-ko rule?
Now that I come to think of it... probably not.
Best,
Chris
Chris Lawrence
> On Wed, 6 Oct 2004 00:26:20 +0100, Chris Lawrence
> <new...@holosys.co.uk.invalid> wrote:
> >You seem to think that rules
> >are everything, in every aspect of your psyche as far as I can see.
>
> When lacking factual arguments, change topic by discussing psyche and
> moral
Don't you dare reposition my comments that way. I'm simply saying that
you keep going on about something which, to the vast majority of people
that I can see, is a non-issue. I'm trying to understand why you have
an obsession with rules and labels and what it brings to your Go raison
d'être.
> >it means that you're like a obsessive legal pedant who nitpicks
> >endlessley over small things
>
> and by being "polite"...
It was neither polite nor not polite, merely a statement of fact.
--
Chris
robert pauli
Thinking again about it, it could
also be that KGS' dispute breaker is
- both tag _all_ opposing stones that seem dead
- should your opponent disagree, play a test
- clear ko bans
- let defender start (two tests in all)
- a pass clears _one_ ko ban, to be chosen
- three passes in a row end test
- if stones of the defender sit in the
tagged area, _none__ of his stones was dead
This would prevent Black in
---------------------------
| X X X . . X . X . . X X X |
| O O O O O X X X O O O O O |
---------------------------
from tagging _all_ white stones as dead,
however, Black could tag only one group:
Again not the real thing (and BTW not
deterministic).
Best
RP
robert pauli
message news:<41629b6...@news.snafu.de>...
>> Also, most people would not like it when
>> they had to fill all dame.
> This seems to be the major point of Japanese-style
> rules on go servers. It does not require the
> Logical Japanese Rules of Go with their side-effects
> that are much more frequent than those of NAJ with
> the above modifications.
Dear Robert,
please give me an example of LJRG's side-effect
to let me guess how frequent it would be and how
non-Japanese.
TIA and Best
RP
robert pauli
>>> New Amateur-Japanese Rules,
>> Faintly remember he continued from the "final"
>> position twice, each color starting once.
>> This may define a result, but I still don't
>> get how this categorizes stones and space in
>> the "final" position. We have to ask RJ.
> What are your questions?
Hi Robert.
The first question just would be if Brian is
right in that NAJ would treat 3PWC as 3 points
in favor of the lone stone's owner, that's,
is this
-------------
5 | . . . . X . |
4 | X X X X X O |
3 | X O O O O O |
2 | O X X O . O |
1 | . X X O O . |
-------------
a b c d e f
7 - 2 = B+5 under NAJ ?
The second question would be, why exactly?
For the third I first have to look at NAJ
again . . . (
- "Two intersections can be adjacent along a line"
defines nothing - told you already.
- You miss to define when _intersections_ connect
(Nit, OC, but aren't you _the_ nitpicker? :-).
- "The position is the distribution of black, white,
and no stones..." - come on, "and no stones"... :-b
- fixed-ko, hmm, A,B,....,A and now not B
= one may not restart a just ended cycle.
Hmm, triple ko, after 3 flips (6 moves) we're back
again, but now I have to insert a ko threat . . .
Intentionally not a void game or tie ??
. . . )
OK, now how would NAJ categorize stones and
intersections in, say, this final position:
-------------
3 | . O X . X . |
2 | O O O O X X |
1 | . O X . X . |
-------------
a b c d e f
OC, White controls a1-a3, b1-b3, c2, and d2,
and Black e1-e3 and f1-f3, but who the heck
controls the rest: c1, c3, d1, and d3 ?
As far I see, this is _not_ defined by NAJ.
Best
RP
PS1
Simply put it as
A player "controls" an intersection in the
final position if he can't be prevented to
occupy it or all its adjacent intersections
with his immortal stone(s) - even if his
opponent starts.
and the problem is gone. :-)
PS2
To all:
Don't confuse NAJ's (naive) "control" with LJRG's.
LJRG won't grant Black control of a single point in
---------------------------
| X X X . . X . X . . X X X |
| O O O O O X X X O O O O O |
---------------------------
whereas NAJ grants him part of the board
(just which? :-) or the whole board, depending
on if NAJ is left as is or made deterministic
as I suggested.
robert pauli
wrote in message news:<cjua1g$dji$1...@rl0001.unimaas.nl>...
> PS How would you score the simplified
> position below?
> ---------------------------
> | . . . O X . X . X O . . . |
> | O O O O X X X X X O O O O |
> ---------------------------
Dear Erik,
case I'm adressed, LJRG scores it as 2 - 0 = B+2.
Its dispute breaker OC too. Why?
First, Black won't tag any white stones as dead
because he couldn't succeed in the test:
Say Black tags the left group as dead and White
disagrees. After Black adds the "bubbles" at f2
and h2, we have (* = permanent white stones)
---------------------------
| . . . O X . X . X * * * * |
| O O O O X X X X X * * * * |
---------------------------
White starts and makes two eyes: Black failed
to cover his claim (beyond *) with a two-eye
formation. (Just the same if Black tags both.)
Second, Black will tag both white "territories"
as neutral because White couldn't succeed in the
test either (* = now permanent black stones)
---------------------------
| . . . O * * * * * * * * * |
| O O O O * * * * * * * * * |
---------------------------
Black starts, kicks'em off, and intrudes permanently.
Best
RP
Robert Jasiek
wrote:
>please give me an example of LJRG's side-effect
>to let me guess how frequent it would be and how
>non-Japanese.
position during the alternation
. # # # # # # # # # #
# # # # # # # O O # #
. # # # # # O . . O #
# # # # # # O O O O #
O O # # # # O . O # #
. O O O O . O O O # #
. O O O O # # # # # #
. # O O O # # # O O #
# . # O O # O O O . O
. # O . O . O . . O O
Different dame have different degrees of importance under LJRG. White
wants to connect the upper dame rather than the lower dame.
Such kos and ko threats are not frequent but also not rare. IMO, more
frequent than bent-4 plus seki elsewhere (which trouble NAJ with only
one analysis) and much more frequent than 1000-year ko plus seki
elsewhere (which trouble NAJ). J2003 does not have problems with the
above position (the white ko stone is capturable-1).
--
robert jasiek
Robert Jasiek
>. # # # # # # # # # #
># # # # # # # O O # #
>. # # # # # O . . O #
># # # # # # O O O O #
>O O # # # # O . O # #
>. O O O O . O O O # #
>. O O O O # # # # # #
>. # O O O # # # O O #
># . # O O # O O O . O
>. # O . O . O . . O O
>J2003 does not have problems with the
>above position (the white ko stone is capturable-1).
Oops, mistake. The stone is capturable-2:
Black captures ko, White approaches from other side, Black defends other
side, White has to kopass, Black connects. I.e. White does not get a C1
stone, only a C2 stone. Anyway, this means that the white ko stone is
alive and for NAJ/LJRG this has some consequences, as discussed below.
(Under J1989 the position is too unclear with respect to enabling.)
O O # . # # .
. O # # # # #
. O O O # O O
. # O O O . .
# . # O O O O
. # O . O O .
NAJ:
white-analysis:
O O # . # # .
. O # # # # #
2 O O O # O O
3 # O O O 5 7
# 6 # O O O O
8 # 4 1 O O .
9, 10, 11 = pass.
So NAJ behaves like LJRG here. Interesting;) This means that 1000-year
ko plus seki elsewhere is not the most frequent tradition breaker for
NAJ but positions like the above. Now NAJ and LJRG seem equally bad or
good as far as modelling tradition is concerned.
--
robert jasiek
Robert Jasiek
wrote:
>The first question just would be if Brian is
>right in that NAJ would treat 3PWC as 3 points
>in favor of the lone stone's owner, that's,
>is this
>
> -------------
> 5 | . . . . X . |
> 4 | X X X X X O |
> 3 | X O O O O O |
> 2 | O X X O . O |
> 1 | . X X O O . |
> -------------
> a b c d e f
>
>7 - 2 = B+5 under NAJ ?
Under NAJ, it is recommended to fill all dame during the alternation, so
let me assume that the dame was filled and I do not need to consider
dame being controlled by some player or another.
Then you have introduced a devilish ko threat. If you want to understand
3PWC, then let us study its basics first before adding complications.
So let us have this
final-position
. . # . # .
# # # # # #
# O O O O O
O # # O . O
. # # O O .
black-analysis
10. # . # .
# # # # # #
# O O O O O
3 4 9 O . O
6 5 7 O O .
8 at 3.
pass = 1, 11, 12, 13.
position at the end of the black-analysis
# . # . # .
# # # # # #
# O O O O O
# # O O . O
# O O O O .
Black control B in the final-position
B B B B B B
B B B B B B
B O O O O O
B # # O . O
B B # O O .
black-score = 7
***
white-analysis
. . # . # .
# # # # # #
# O O O O O
8 2 5 O . O
1 3 4 O O .
6 at 1, 7 at 3, 9 at 4.
pass = 10, 11, 12.
position at the end of the white-analysis
. . # . # .
# # # # # #
# O O O O O
# O O O . O
# # O O O .
White control W in the final-position
. . # . # .
# # # # # #
# W W W W W
O W W W W W
. # W W W W
white-score = 8
***
score = 7 - 8 = -1
***
Scores within the 3PWC shape:
*1 is for empty intersections,
*2 is for occupied intersections.
black-local-score = 1 * 1 + 1 * 2 = 1 + 2 = 3
white-local-score = 0 * 1 + 3 * 2 = 0 + 6 = 6
local-score = 3 - 6 = -3
White scores 3 points in the 3PWC.
>The second question would be, why exactly?
The players player 8we assume) optimally in the black-analysis and then
in the white-analysis. An example of such perfect play analysis
sequences is shown above. Is that clear enough or do I need to bury it
under the rules terms "control" and "immortal"?
>For the third I first have to look at NAJ
>again . . . (
>- "Two intersections can be adjacent along a line"
> defines nothing - told you already.
>- You miss to define when _intersections_ connect
> (Nit, OC, but aren't you _the_ nitpicker? :-).
>- "The position is the distribution of black, white,
> and no stones..." - come on, "and no stones"... :-b
AYK, the rules are a semi-formal text, not a mathematically formal text.
For the latter, your objections would be useful.
>- fixed-ko, hmm, A,B,....,A and now not B
> = one may not restart a just ended cycle.
Wrong. One may not restart a just ended cycle WITH THE SAME PLAY.
> Hmm, triple ko, after 3 flips (6 moves) we're back
> again, but now I have to insert a ko threat . . .
During the alternation or an analysis? Anyway, triple ko is discussed
exhaustively on the commentary webpage.
> Intentionally not a void game or tie ??
>. . . )
Exactly.
>OK, now how would NAJ categorize stones and
>intersections in, say, this final position:
Stones and intersections are categorized as to whether they score for a
player.
> -------------
> 3 | . O X . X . |
> 2 | O O O O X X |
> 1 | . O X . X . |
> -------------
> a b c d e f
>
>OC, White controls a1-a3, b1-b3, c2, and d2,
>and Black e1-e3 and f1-f3, but who the heck
>controls the rest: c1, c3, d1, and d3 ?
>
>As far I see, this is _not_ defined by NAJ.
Control does not exist for the final-position without knowing the
analyses, but one has to look at the position at the end of the
black-analysis and the position at the end of the white-analysis. To
find out, both analyses have to be known, i.e. in practice the
[disputing] players have to execute them.
final-position
. O # . # .
O O O O # #
. O # . # .
black-analysis
. O 3 1 # .
O O O O # #
. O # 2 # .
4, 5, 6 = pass
position at the end of the black-analysis
. O O O # .
O O O O # #
. O # # # .
Black control B in the final-position
. O # . B B
O O O O B B
. O B B B B
black-score = 3
***
white-analysis
. O # 1 # .
O O O O # #
. O 4 2 # .
pass = 3, 5, 6, 7
position at the end of the white-analsis
. O # # # .
O O O O # #
. O O O # .
White control W in the final-position
W W # . # .
W W W W # #
W W W W # .
white-score = 5
(In the final-position, W controls 1 occupied intersection with a black
stone and 3 empty intersections. This gives 5 points.)
***
score = 3 - 5 = -2
> A player "controls" an intersection in the
> final position if he can't be prevented to
> occupy it or all its adjacent intersections
> with his immortal stone(s) - even if his
> opponent starts.
>
>and the problem is gone. :-)
Problem? Which?
>whereas NAJ grants him part of the board
>(just which? :-) or the whole board, depending
>on if NAJ is left as is or made deterministic
>as I suggested.
If you mean J2003-deterministic in the "prevent",
then that would be contrary to NAJ's intention
since one would have to include h-strategy in
the NAJ rules.
--
robert jasiek
Robert Jasiek
position at the end of the white-analysis
. . # . # .
# # # # # #
# O O O O O
# O O O . O
# # O O O .
White control W in the final-position
. . # . # .
# # # # # #
# W W W W W
O W W W W W
. # W W W W
<<
Correction:
position at the end of the white-analysis
. . # . # .
# # # # # #
# O O O O O
# # O O . O
# O O O O .
White control W in the final-position
. . # . # .
# # # # # #
# W W W W W
O # W W W W
. W W W W W
***
Concerning frequency of positions that in hypothetical-analysis differ
under a basic-ko rule and some ko-pass rule(s):
final-position
O O # . # .
. O # # # #
. O O O O O
. # O O O .
# . # O O O
. # O . O .
intersections
O O # . # .
. O # # # #
a O O O O O
b # O O O .
# d # O O O
. # e f O .
Status of white ko stone?
Basic-ko rule:
#[fabepfpp]
"C1"
Ko-pass rule:
#[fabPepp]
"not C1"
***
The white ko stone is C2 but not C3 under J2003. So for J2003, this
shape provides a third kind of C2 strings, although it is related to
bent-5.
--
robert jasiek
robert pauli
Dear Robert,
thanks for the answers (and
don't forget the other...).
So, NAJ would categorize this
-------------
5 | . . X . X . |
4 | X X X X X X |
3 | X O O O O O |
2 | O X X O . O |
1 | . X X O O . |
-------------
a b c d e f
as
-------------
5 | X X X X X X |
4 | X X X X X X |
3 | X O O O O O |
2 | X X O O O O |
1 | X O O O O O |
-------------
a b c d e f
producing three extra points for Black
and six extra for White in the lower
left corner: W+1.
Funny to see just three of four connected
stones treated as dead, but anyway, that's
not really the problem I have with NAJ.
It's that there are final positions where,
even if we assume perfect play in the
two-fold analysis, NAJ's categorization
is _NOT_ defined.
I gave you an simple example - here's an
even simpler one:
-----------
4 | . O X X . |
3 | . O . X . |
2 | . O . X . |
1 | . O O X . |
-----------
a b c d e
Who controls which dame in the center?
BA1: c3/c2... = B controls c2
WA1: c2 c3... = W controls c3
BA2: c2/c3... = B controls c3
WA2: c3 c2... = W controls c2
BA3: c3/c2... = B controls c2
WA3: c3 c2... = W controls c2
(and nobody controls c3)
BA4: c2/c3... = B controls c3
WA4: c2 c3... = W controls c3
(and nobody controls c2)
Now come on, Robert, are you selling
me this as a "definition" ??
If you rather do it as I suggest, that's
A player "controls" an intersection in the
final position if he can't be prevented to
occupy it or all its adjacent intersections
with his immortal stone(s) - even if his
opponent starts.
there would only be _one_ categorization:
nobody controls c2
nobody controls c3
(pretty sound, ehh :-)
Each intersection now has it's own two-fold
analysis, but that won't bother amateurs a
bit, believe me, since they'll skip the fuss
anyway. Note that this is much safer than
skipping the two-fold whole-board analysis!
Best
RP
Robert Jasiek
wrote:
>It's that there are final positions where,
>even if we assume perfect play in the
>two-fold analysis, NAJ's categorization
>is _NOT_ defined.
Yes, the categorization is done when the players have chosen the
analyses, not before. This is intended.
>Now come on, Robert, are you selling
>me this as a "definition" ??
It depends on your definition of "definition". If you do not allow
variables, then it is not a definition, if you allow variables, then it
is a definition.
>Each intersection now has it's own two-fold
>analysis, but that won't bother amateurs a
>bit, believe me,
Rules ought not to deceive players. You seem to argue contrarily:
hypothetical-strategy shall be hidden in the rules to deceive players.
>since they'll skip the fuss
>anyway. Note that this is much safer than
>skipping the two-fold whole-board analysis!
I don't follow.
--
robert jasiek
robert pauli
>> please give me an example of LJRG's side-effect
>> to let me guess how frequent it would be and how
>> non-Japanese.
> position during the alternation
>
> -----------------------
> | . # # # # # # # # # # |
> | # # # # # # # O O # # |
> | . # # # # # O . . O # |
> | # # # # # # O O O O # |
> | O O # # # # O . O # # |
> | . O O O O . O O O # # |
> | . O O O O # # # # # # |
> | . # O O O # # # O O # |
> | # . # O O # O O O . O |
> | . # O . O . O . . O O |
> -----------------------
>
> Different dame have different degrees of importance
> under LJRG. White wants to connect the upper dame
> rather than the lower dame.
OK, thanks, I see: You don't like local threats
being effective under LJRG.
For the rest:
- connecting up adds _no_ local threat because
only the lower eye will be used when proving
control of the lower left corner, whereas
- connecting down adds a local threat because
you also have to use the three stones to let
your proof benefit from any liberties there.
However, Robert, you miss a fine point:
Even _if_ White connects up, she won't spare a
teire! Why? Because just before end Black will
convert the ko to a direct one:
| 2 O O O O # # # # # # |
| 3 # O O O # # # O O # |
| # . # O O # O O O . O |
| . # O 1 O # O . . O O |
-----------------------
and 4 takes back. No matter how many ko threats
White has, she can't leave it unfilled because
she couldn't prove control (Black starts...):
bingo.
Not really a good try, Robert - just serving me
back my very own example (in very bad shap, BTW).
Any better one, buddy, to keep your claim up ?
Best
RP
Robert Jasiek
wrote:
>> -----------------------
>> | . # # # # # # # # # # |
>> | # # # # # # # O O # # |
>> | . # # # # # O . . O # |
>> | # # # # # # O O O O # |
>> | O O # # # # O . O # # |
>> | . O O O O . O O O # # |
>> | . O O O O # # # # # # |
>> | . # O O O # # # O O # |
>> | # . # O O # O O O . O |
>> | . # O . O . O . . O O |
>> -----------------------
>Any better one, buddy, to keep your claim up ?
Ok, then you should not use unfilled dame examples for NAJ since all
non-seki dame can be filled during the alternation in perfect play.
-----------------------
| . # # # # # # # # # # |
| # . # # # # # O O # # |
| # # # # # # O . . O # |
| O O # # # # O O O O # |
| # O # # # # O . O # # |
| . O O O O . O O O # # |
| # O O O O # # # # # # |
| . # O O O # # # O O # |
| # . # O O # O O O . O |
| . # O . O . O . . O O |
-----------------------
Black creates a direct ko again. However, this time the corner shape is
from you:) I do not find better examples now but have another question:
Have you transformed LJRG so that only one analysis for all strings
together is needed?
--
robert jasiek
robert pauli
> Ok, then you should not use unfilled dame examples
> for NAJ since all non-seki dame can be filled during
> the alternation in perfect play.
Had no intention to fool NAJ or you. Dame is useless
under LJRG, and I'm so used to it that I forget that
some rulesets, even if the call themselves "Japanese",
might let them have an effect.
"Your" second try:
> -----------------------
> | . # # # # # # # # # # |
> | # . # # # # # O O # # |
> | # # # # # # O . . O # |
> | O O # # # # O O O O # |
> | # O # # # # O . O # # |
> | . O O O O . O O O # # |
> | # O O O O # # # # # # |
> | . # O O O # # # O O # |
> | # . # O O # O O O . O |
> | . # O . O . O . . O O |
> -----------------------
> Black creates a direct ko again.
> However, this time the corner shape is from you:)
But only the corner shape :-)
Just can't bear to see a single 2x2
block - not to mention a 4x5 =:-O
> I do not find better examples now...
Here's my new one, buddy (only to be
copied while stating its inventor :-)
-------------------
9 | . O X X . X . X O |
8 | . O O X . X . X O |
7 | . . . O X X X O O |
6 | O O O O O X O . . |
5 | X X . O X . O . O |
4 | . X O O X . O . . |
3 | . X O X . X O O . |
2 | O X O X . X X O . |
1 | X . X X . . X O . |
-------------------
a b c d e f g h i
Black's turn. Connect at f5 or f4? Normally it
doens't matter, but under LJRG it does!
For all:
f5: If White takes f4, Black has to add a stone
in the lower left corner, losing by one.
Why? Because "locally" there are two threats
at e8 and g8, letting White tag the bubble
at b1 as neutral and disagree if Black tags
the white stone at a2 as dead.
("Local" is _no_ LJRG term, but Black needs
the liberties at e3, therefore he needs e4,
therefore he has to include e8 and g8.)
f4: Whoever takes f5, Black can now spare a stone
in the lower left corner due to two ko threats
on the right side - and after end there are no
local threats for White.
Is this my intention? Yes! I easily could change
LJRG to use "pass for ko" (for one or all kos) in
the definition of control, but why introduce an
exception? To stick to a tradition that probably
is nothing but an over-simplification? No, sir.
> ...but have another question:
> Have you transformed LJRG so that only one
> analysis for all strings together is needed?
No (see http://de.geocities.com/bobla62/ljrg.html),
but its dispute breaker nevertheless does with
at most _one_ test - you see (no O(...) fuss :-).
Best
RP
robert pauli
> Rules ought not to deceive players.
> You seem to argue contrarily:
> hypothetical-strategy shall be hidden
> in the rules to deceive players.
Come on, Robert. First you tell me NAJ is
semi-formal, and now you tell me I'm skipping
to formalize "can't prevent" in them.
>> Note that this is much safer than skipping
>> the two-fold whole-board analysis!
> I don't follow.
Don't want to, or not able to ?
Anyway, what I mean is that if the categorization
is defined by the final position, as it should be,
skipping the fuss of formally check it is possible
and likely.
But if the final position is not defining the
categorization in all cases, if the game actually
has to be continued twice, each color starting once,
then we only can skip it if it doesn't affect the
result. This is not so easy to agree on as about
who controls a single intersection in the case
above, IMHO.
Best
RP
Robert Jasiek
>(see http://de.geocities.com/bobla62/ljrg.html),
>but its dispute breaker nevertheless does with
>at most _one_ test
Where in your page should I look to find the dispute
breaker?
--
robert jasiek
Robert Jasiek
wrote:
>First you tell me NAJ is
>semi-formal, and now you tell me I'm skipping
>to formalize "can't prevent" in them.
Sure. "semi-formal" refers to the degree of formalism of each definition
text's wording. However, I have not meant it to refer to completeness of
each definition text's contents. NAJ does not use non-trivial terms that
are not defined. "prevent" would be a non-trivial, non-defined term.
Consider NAJ's text for "Black controls":
"In the final-position, Black controls an intersection if the
black-analysis ends with immortal stones of Black on the intersection or
on all its adjacent intersections."
The following words are terms and are (roughly) defined or used like
axioms in the rules:
final-position
Black
controls
intersection
black-analysis
immortal
stones
adjacent
The following words are are not defined but well understood part of the
English language:
in
the
an
if
ends with
of
on
or
all
its
Now no further word remains. You want to introduce another word / term:
"prevent". This may be part of the English language, however, it is a
mighty word, especially as a go rules term. Mighty words used in
semi-formal go rules must be defined and not treated like articles,
pronouns, conjunctions, or prepositions.
--
robert jasiek
robert pauli
>> (see http://de.geocities.com/bobla62/ljrg.html),
>> but its dispute breaker nevertheless does with
>> at most _one_ test
> Where in your page should I look to find the
> dispute breaker?
Klick the darn link (short)
and read the last point (17).
Best
RP