Fixed Ko Rule: Continued Discussion

Magicwand · **#21**

hyperpape wrote:

This is just curiousity, I know it won't sway your interest in these positions, but has such a very difficult case arisen in play? I don't mean Herman's sort of artificial example, but any long-cycle? More than a triple-ko, I guess?

i have seen quadriple ko few times.
even triple ko is hard to apply superko rule.
imagine 4 or 5.

super ko will solve one probem but also will bring many more problems.
i do not understand why people try to rid of no decision rule.
it rarely happens and playing another game should be easy solution.

HermanHiddema · **#22**

hyperpape wrote:

This is just curiousity, I know it won't sway your interest in these positions, but has such a very difficult case arisen in play? I don't mean Herman's sort of artificial example, but any long-cycle? More than a triple-ko, I guess?

Quadruple ko is certainly not unheard of (link gives several examples of them happening in actual play). John Fairbairn mentions at the Triple Ko page that GoGoD has 10 quadruple ko examples and mentions having heard of a quintuple ko in a Japanese amateur game.

Other strange cycles that arose in play include: Sending Three Returning Two (in one of my own games) and Molasses Ko (in a game of T Mark Hall).

hyperpape · **#23**

Thanks, all.

RobertJasiek · **#24**

Magicwand wrote:

even triple ko is hard to apply superko rule.

Ordinary triple ko is easy to apply under superko. Well, if you have not through about it yet, do so and you will understand the why.

Quote:

super ko will solve one probem

Like every ko ruleset, superko solves the problem of meaningless endlessly recurring cycles. This is much more than your understatement of [only] "one problem" suggests.

Quote:

but also will bring many more problems.

Difficult strategy you call problems? If so, yes.

Quote:

i do not understand why people try to rid of no decision rule.

Because it is an extremely bad rule with gaps and ambiguity:
http://home.snafu.de/jasiek/mistakes.html
http://home.snafu.de/jasiek/j1989c.html
Read the relevant parts of those pages incl. the commentary on rule §12 and don't miss the Interpretative Text! If one spells out No Result Rules properly, then their text becomes more difficult than the other mentioned ko rulesets.

Quote:

it rarely happens

If rarity is your concern, then you should have nothing against superko! Any strategic difficulties are just (very) rare. So why care?

Quote:

and playing another game should be easy solution.

One cannot say so in general because of limited tournament schedules. A more generally working solution would be to say that Black (alternatively: White) wins in case of a long cycle.

HermanHiddema · **#25**

Another solution to long cycles: Declare all cycle intersections off limits.
Remove any stones from them and prohibit both players from playing there.

This would apply to all cycles longer than 2 (or longer than 3, if you prefer).

Examples:

Click Here To Show Diagram Code: [go]$$B Triple ko: Before $$ --------------- $$ . X O . O X O . $$ . X O O X X O . $$ . X O . O X O . $$ . X O O X X O . $$ . X O X . X O . $$ . X O O X X O . $$ . X X X O O O . $$ . . . . . . . .[/go]

If the players play through the cycle (6 moves), then:

Click Here To Show Diagram Code: [go]$$B Triple ko: After $$ --------------- $$ . X O C C X O . $$ . X O O X X O . $$ . X O C C X O . $$ . X O O X X O . $$ . X O C C X O . $$ . X O O X X O . $$ . X X X O O O . $$ . . . . . . . .[/go]

All the

marker intersections are now cleared and prohibited. Effectively, the result is seki.

Second example. Eternal life:

Click Here To Show Diagram Code: [go]$$B Eternal life: Before $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | O O O O O O . . . . $$ | X X X X X O . . . . $$ | X O O O X X O O . . $$ | . O . X . O X . . . $$ --------------------[/go]

If the player play through the (four move) cycle, then:

Click Here To Show Diagram Code: [go]$$B Eternal life: After $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | O O O O O O . . . . $$ | X X X X X O . . . . $$ | X O O O X X O O . . $$ | . O C C C C X a . . $$ --------------------[/go]

Again, the :ec:

marked points are cleared and prohibited. Again the result is effectively seki.

(Note that in this case, optimal strategy for White would normally be to play at a on move 4, rather than throwing in, because the result is also seki, but with White getting sente and a move on the outside.)

With this rule, the result of long cycles will simply be seki. The players can continue with the game and get a result through play, which is IMO a much preferred solution to any declared result (such as "no result", "draw" or "white/black wins").

You could, of course, still make a distinction between balanced (same number of captures) and unbalanced (different number of captures) cycles, like long cycle rules do, and declare a winner in the unbalanced cases.

RobertJasiek · **#26**

Removing stones with liberties defies the simple beauty of the capture rule. To work out Robert Pauli's / your proposal a bit differently, I suggest to leave the stones on the board but to prohibit playing on any earlier played cycle set intersection. Let me call this the Cycle Set Prohibition Rule.

HermanHiddema · **#27**

RobertJasiek wrote:

Removing stones with liberties defies the simple beauty of the capture rule. To work out Robert Pauli's / your proposal a bit differently, I suggest to leave the stones on the board but to prohibit playing on any earlier played cycle set intersection. Let me call this the Cycle Set Prohibition Rule.

This has very different consequences. The rule I suggested effectively results in seki. With this rule, there can be a "winner" of the cycle, with somewhat strange prohibited points remaining.

E.g:

Click Here To Show Diagram Code: [go]$$W Eternal life: Before $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | O O O O O O . . . . $$ | X X X X X O . . . . $$ | X O O O X X O O . . $$ | . O X X . O X . . . $$ --------------------[/go]

Click Here To Show Diagram Code: [go]$$W First two moves $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | O O O O O O . . . . $$ | X X X X X O . . . . $$ | X O O O X X O O . . $$ | . O . 2 1 O X . . . $$ --------------------[/go]

Click Here To Show Diagram Code: [go]$$W Next two moves $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | O O O O O O . . . . $$ | X X X X X O . . . . $$ | X O O O X X O O . . $$ | b O 4 X a 3 X c . . $$ --------------------[/go]

Now, the players are prohibited from playing at a, but Black can capture at b, and White at c resulting in:

Click Here To Show Diagram Code: [go]$$W Eternal life: After $$ | . . . . . . . . . . $$ | . . . . . . . . . . $$ | O O O O O O . . . . $$ | X X X X X O . . . . $$ | X . . . X X O O . . $$ | X . X X a O . O . . $$ --------------------[/go]

This position, where a is prohibited to both players.

Effectively, white has lost the eternal life fight here.

Click Here To Show Diagram Code: [go]$$B $$ --------------------------------------- $$ | . O X . X O . . . . . . . O X a X X O | $$ | O O O X X O . . . . . . . O X X X O . | $$ | . O X X X O . . . . . . . O O X O O O | $$ | O X X . X O . . . . . . . . O O X O . | $$ | X X X X X O . . . . . . . O . O X O O | $$ | O O O O O O . . . . . . . . . X X X X | $$ | . . . . . . . . . . . . . . . . . . . |[/go]

In a similar way, a triple ko consisting of double ko seki and a another ko may in this may result in a favorable result for one player, where the double ko becomes prohibited while one player wins the other ko.

In the above position, if the ko is taken, then the double ko exchanged, then the ko taken back, then the double ko exchanged again, white will "win" the ko, because he is allowed to take at a later. The empty 2-1 point will be prohibited.

Harleqin · **#28**

I do not think that arbitrary rule amendments can give a satisfactory solution. I believe that a practical ruleset should have a single ko rule that covers all corner cases.

"Positional superko", "situational superko", "natural situational superko", and "double fixed ko"(*) seem to fulfill this requirement. "Basic ko + long cycle tie", "basic ko + fixed ko" do not, while "Basic ko + mushoubu" is a question of perspective.

(*) This has not yet been studied much. It says (roughly) that a move, characterized only by the (unordered) set of its before- and after-position, may not be repeated. I believe it was proposed in some thread either here or on the old godiscussions.com forum.

HermanHiddema · **#29**

Harleqin wrote:

I do not think that arbitrary rule amendments can give a satisfactory solution. I believe that a practical ruleset should have a single ko rule that covers all corner cases.

What is wrong with having a set of rules, rather than a single rule, if the set covers all corner cases?

I do think it is bad to have a single rule that does not cover all cases, and to then add rulings on a case by case basis. But I see nothing wrong with rules that treat basic ko different from other cycles, as long as they cover all cases.

RobertJasiek · **#30**

"wrong" would be too strong a word but "disadvantage" fits:

More than one rule makes reading the text, understanding the text as such and thus learning the rules as a newbie / journalist / other interested non-player a bit more difficult. Possibly good layout for a complete single page rules flyer is made harder.

nagano · **#31**

Apologies, I missed this thread at first. The main question is what is the simplest solution for an all-encompassing ko rule while still allowing kos to be contested and decided by the players' skill and the strength of their positions. Situational superko seems the best solution presently, but it has flaws in some positions. I will continue this discussion later but need to go presently.

RobertJasiek · **#32**

Harleqin, quite like cycle stone removals, doubled fixed ko rule is an artificial construction: It even prohibits something that cannot be executed as a play. Therefore the rule should not be used in human play.

RobertJasiek · **#33**

nagano wrote:

what is the simplest solution for an all-encompassing ko rule while still allowing kos to be contested and decided by the players' skill and the strength of their positions. Situational superko seems the best solution

Positional superko is simpler, so, by your standard of wishing the simplest solution, situational superko does not qualify.

nagano · **#34**

RobertJasiek wrote:

Positional superko is simpler, so, by your standard of wishing the simplest solution, situational superko does not qualify.

Perhaps I should rephrase my goal as "simplest without compromising the potential for variation". This was part of the reason I objected to the fixed ko rule. I object to the positional ko rule as well because of situations like this one. So see it does meet my standards, at least in that regard.

RobertJasiek · **#35**

Quote:

my goal as "simplest without compromising the potential for variation". This was part of the reason I objected to the fixed ko rule.

Your goal is still too imprecise. Which variation do you mean? If you mean number of legal move-sequences, then the fixed ko rule used alone beats superko very easily. Therefore you must mean a different kind of variation but which? Maybe something related to strategic choice complexity? Such is extraordinarily hard to assess though! E.g., PSK has shapes (like Molasses Ko) that are strategically much more demanding than under SSK or NSK. So you will have a very hard time arguing that PSK would be compromising in strategic choice complexity. Rather it is easier to argue that among those ko rulesets with in practice comparable strategic choice complexity PSK is the simplest.

Why at all do you object to a particular ko rule on the grounds of a particular shape's slightly surprising behaviour?

Which (subjective) objection do you have with respect to your linked example? Why do you expect Black to play 2?

Harleqin · **#36**

RobertJasiek wrote:

Harleqin, quite like cycle stone removals, doubled fixed ko rule is an artificial construction: It even prohibits something that cannot be executed as a play. Therefore the rule should not be used in human play.

I think that you have not understood what I have written (that might be my fault). I see no base for your claim; it is not any more artificial than superko. What prohibition do you mean?

If you see the game as a graph of edges (moves) and vertices (positions), superko prohibits revisit of a vertex, while double fixed ko prohibits revisit of an edge (edges are treated as being undirected for this). Both have the same result as basic ko in the basic case, but they differ in the long case (I think that double fixed ko roughly treats each long cycle as something like a "disturbing ko").

willemien · **#37**

Harleqin wrote:

If you see the game as a graph of edges (moves) and vertices (positions), superko prohibits revisit of a vertex, while double fixed ko prohibits revisit of an edge (edges are treated as being undirected for this). Both have the same result as basic ko in the basic case, but they differ in the long case (I think that double fixed ko roughly treats each long cycle as something like a "disturbing ko").

Thanks for the explaination i lost the meaning of the "double fixed ko rule".

It has some strange concequences

if white has played position A -> White move -> position B
then Black may not play position B -> Black move -> position A

Is that really what you want?

HermanHiddema · **#38**

willemien wrote:

Harleqin wrote:

If you see the game as a graph of edges (moves) and vertices (positions), superko prohibits revisit of a vertex, while double fixed ko prohibits revisit of an edge (edges are treated as being undirected for this). Both have the same result as basic ko in the basic case, but they differ in the long case (I think that double fixed ko roughly treats each long cycle as something like a "disturbing ko").

Thanks for the explaination i lost the meaning of the "double fixed ko rule".

It has some strange concequences

if white has played position A -> White move -> position B
then Black may not play position B -> Black move -> position A

Is that really what you want?

You realize that the sequence (A -> B -> A) you describe is exactly equivalent to basic ko?

nagano · **#39**

After being puzzled by your last post, I looked at the definition again. I noticed in the notes that it refers to the whole board position, not local as I had read it. I therefore withdraw my objection.

RobertJasiek · **#40**

Harleqin, double fixed ko rule: Suppose a play by Black captures at least 2 stones, transforming position A to position B. Afterwards neither player, in particular White, may make a play that transforms B to A. However, such a move does not even exist! This is where the rule is artificial.

Superko or the fixed ko rule do not have such artificial aspects of prohibiting plays that cannot even exist in theory.

Fixed Ko Rule: Continued Discussion

Who is online