Life In 19x19

(First draft - trivial definitions omitted, just showing the logical units and how they work together.
See also on Sensei's.)

• 1. Basic rules and trivial definitions (board, intersections, move = board play or pass, strings, liberties, capture - with [3] suicide allowable, resign, handicap)

• 2. Basic ko (if a single stone captures a single stone, it cannot be recaptured immediately on the next move).

• 3. Threefold repetition (draw/loss/win depending on stones captured in the loop). Applied for board plays only and thoughout all phases together.

• 4. Minimal-superko: a board play must not recreate the position of the last successive two passes.

• 5. Three successive passes stop the game and ends a game phase. Resumption from a stop is requestable, starting a new game phase, with intact move order, for a finite times. (First phase is "main game", later phases "dispute phases".)

AREA MODE:

• A.6. A player's score is the number of intersections controlled (occupied or surrounded by his live stones), plus W komi.

• A.7. In a stopped position dead strings are what their owner acknowledges, other strings alive.

TERRITORY MODE WITH PASS STONE PLAYOUT:

• B.6. A player's score is the number of intersections in his territory, plus earlier prisoners taken, plus opponent's stones in his territory ("dead stones"), plus W komi.

• B.7. In a stopped position dead strings are what their owner acknowledges, other strings alive. A player's territory are the intersections that are empty or contain a dead stone of the opponent, and are surrounded by his live stones.

• B.8. In dispute phases a player who passes gives a stone to the opponent as prisoner.

• B.9. In dispute phases the number of moves played by both players must be equal. If necessary, the stopping passes are complemented by an obligatory extra pass of the next player.

TERRITORY MODE WITH HYPOTHETICAL PLAY:

• C.6. A player's score is the number of intersections in his territory, plus earlier prisoners taken, plus opponent's stones in his territory ("dead stones"), plus W komi.

• C.7. {---}

This came up in the other topic, maybe deserves one of its own. The basic idea is that a refinement of the area + superko approach is sufficient to describe the complete Chinese game, including triple ko draws:

• basic ko with a clear threefold repetition rule
  (draw/loss/win depending on stones lost in the loop)
• and clearly stating that superko only applies in dispute phases, after first resumption
  (with these rules pass can lift superko bans)

I don't really understand why Western rules started off by crippling a well established game and its draw on repetition rule. But since it's easy to do otherwise, returning to the Asian roots seems appropriate - even as a first step in faint hope of a future rules unification.

(Logical next step could be territory option with pass stone disputes, followed by pure territory mode or better ko options - more on ko later).

You try to make it an Eastern / Western dispute and pretend consistent Asian tradition but this is not so. In Asia, there have been two factions for a long time: precedental ko rules versus superko. E.g., the first official written rules, the Japanese 1949 Rules, had both a statement of the principle of prohibiting repetition of the whole board position and precedental ko rules. E.g., Chinese Rules tend to have both. Among the Chinese representatives at the International Go Rules Forum, both factions were present.

EDIT: typo factions.

Sounds like a good idea. I don't see how it matters if you were "pretending" or not that there is a west - east divide. Such motivation is hardly something negative in itself and it is anyway a huge leap of the imagination to say that you were.

I don't doubt various rules ideas exist in Asia as well, and even basic things were disputed in the past. But in current practice there seems no problem with long cycles in China (nor Japan or Korea) - only in the West. The majority plays Go assuming triple ko is valid and leads to draw / no result.

It is also much easier to recognize repetition after a few cycles than spotting the first move that recreates a position. So why Western rules need to "know better" - when staying with the majority leads to easier applicable rules as well?

A robust, prisoner-aware threefold repetition rule have further advantages. It allows using superko with passes lifting bans (fixing 1-eye-flaw and similar problems). It also makes suicide allowable even in the basic ko phase - one less extra rule. And for territory scoring, it normalizes hypothetical play (in defending lines with infinite stone losses).

jann wrote: I don't really understand why Western rules started off by crippling a well established game and its draw on repetition rule. But since it's easy to do otherwise, returning to the Asian roots seems appropriate - even as a first step in faint hope of a future rules unification.

Was it Western rules that determined that komi should be set at a fractional value to prevent jigo?
What I am trying to say, is that it is not western influence, per se, at work here, as it is the overall modernization trend of Go to meet the (perceived) needs of orderly competition. This is why there was a perceived need to attempt to actually write down the rules. This drives other innovations as well.

I think perhaps we should not be looking for definitive one-size fits all rule. Specifically, in high stakes, (single game) competition formats, perhaps there should be a definitive rule on Superkos, and the competitors have to work within that rule to win, just to be consistent with requiring a fractional komi to produce a winner. I think in any other case, it should be allowed to produce a draw or NR, and not bother with the rule. (But then I also think komi should always be set at 7.0 because I think that a true draw is beautiful.)

jann wrote:It is also much easier to recognize repetition after a few cycles than spotting the first move that recreates a position.

No-result rules have these aspects:
1) Generous instead of strict recognition of recreation.
2) The special result no-result that is uncomparable to scored results, such as jigo.
3) For territory scoring, the strategically different behaviours of even cycles with the difference zero of their removed stones and of odd cycles with a non-zero difference of their removed stones.

So why Western rules need to "know better" - when staying with the majority leads to easier applicable rules as well?

The majority is beginners and does not know application of whichever rules to long cycles at all.

Japanese, Korean, Chinese (note: not Taiwanese, therefore not in general Asian and Western is not the complement) no-result-like rules are hard to apply because they are confusing, ambiguous, contradictory or incomplete. For ordinary alternation, there should be clarification, such as on my page

http://home.snafu.de/jasiek/long_cycle.pdf

"If a play creates a cycle, then the game's result depends on the numbers of black and white stones removed from the board during the cycle:
a) If equally many black and white stones have been removed, then the result is a tie.
b) If fewer black than white stones have been removed, then the result is a Black win.
c) If fewer white than black stones have been removed, then the result is a White win."

For traditional results, replace "tie" by "no-result" in this text.

Further clarification is required for rules with hypothetical play.

Only after clarification it becomes meaningful to claim "easier applicable".

However, (2) is not easily applicable. Quite contrarily, it leads to undecidable strategy:
http://home.snafu.de/jasiek/mistakes.html
My long-cycle rules linked further above with the result tie avoid at least this fundamental mistake in rules design.

Actually, my 3-plays Ko Rules are simpler and better:

http://home.snafu.de/jasiek/3_plays.pdf

"If a play creates a cycle, then the game ends prematurely. The result is then exceptional, and depends on the number of plays during the cycle:
- Short cycle: If it is 2 or 3, then the player making the last play loses.
- Long cycle: If it is 4 or greater, then the result is a tie."

Such can be accompanied by a tournament rule for (1).

Although I prefer positional superko because I prefer the greatest simplicity of the rules, the two alternatives above and my Basic-Fixed Ko Rules as a third alternative are possible and each easier and simpler than no-result rules.

A robust, prisoner-aware threefold repetition rule

Threefold is unnecessarily complicated. Any number of repetitions of a cycle, including just one cycle, is good enough.

Prisoner-aware is possible but unnecessarily complicated. No-result rules and my Long-cycle rules have it. My 3-plays Ko Rules and Basic-Fixed Ko Rules do not use or need it.

It allows using superko with passes lifting bans (fixing 1-eye-flaw and similar problems).

Where "superko" does not refer to a superko rule.

The other mentioned ko rules also handle all positions.

It also makes suicide allowable even in the basic ko phase - one less extra rule.

Wrong. It is always necessary specify how to execute a play (with or without suicide).

Some ko rules have a side effect on suicide. Superko has the side effect that it automatically prohibits single-stone suicide. Some other ko rules (no-result or long-cycle ko rules) have the side effect that recurring a cycle with a non-zero prisoner difference is strategically wrong under territory scoring. A cycle with suicide of several stones is fun for the researcher.

And for territory scoring, it normalizes hypothetical play (in defending lines with infinite stone losses).

The other ko rules allow handling of hypothetical play regardless of whether such is called normalisation.

RobertJasiek wrote:2) The special result no-result that is uncomparable to scored results, such as jigo.

You are missing the point here. Games can end with equal scores, and can also end on repetition. These are two things thus have different names (jigo and mushoubu, but any term is usable, X and Y).

It is up to the tournament or the environment in which the game was played to decide how to treat these outcomes. In simplest cases they can get equal treatment (like 0.5 pts). In other cases the organizers may decide to have either - or both - replayed. This is not decidable in advance, not in the rules of play (though recommending identical treatment seems fine).

Threefold is unnecessarily complicated. Any number of repetitions of a cycle, including just one cycle, is good enough.

Even beyond the ease of noticeability, there are reasons why threefold repetition is usually required (which is AT LEAST threefold, so also declarable later, as the players realize).

After the first cycle the players may play differently since they now got further information about the opponent's choices. If a single cycle would already end, you would distort the game unnecessarily as near the end of the forming cycle a player may be forced to choose suboptimal moves just to avoid a premature end - instead of simply playing differently the second time.

While tournament rules can override the rules of play, their original no-result design has been intentionally created to be different from the result tie. Therefore my point of undecidable strategy holds, except for tournaments with different tournament rules declaring the result tie.

It may be a good idea to state in rules a default resolution of all the outcomes of the game. The laws of chess for example state if not otherwise specified that a win is 1 pt., draw 0.5 pt. and loss 0 pt.. One could state the same that if not otherwise specified that a win in go is 1 pt., jigo 0.5 pt., a loss 0 pt. and that no result are 0.5 pt (or need to be played again).

What I think you mean by "undecidable strategy" is that "the problem of finding a strategy is not well-defined". I think that is even a bit unprecise because it is really just the "no result" label doesn't have a well-defined value. Please clarify if you mean something completely different.

Strategy is defined, e.g., by me here:

http://home.snafu.de/jasiek/ko.pdf
http://home.snafu.de/jasiek/j2003.html

We can assign the value "no result" to the game result of a "no result" game.

The problem is not the value as such but the problem is that it is uncomparable to any score (in particular the score 0). Therefore, positions like those at

http://home.snafu.de/jasiek/mistakes.html

have undecidable strategy.

(In a last round tournament game, I found myself in a situation of undecidable strategy for a different reason: Should I choose a safe jigo, trick play to possibly win, a complex strategy with the likely result jigo or should I optimise my expected growth of qualification points? I went for "complex strategy with the likely result jigo" and achieved a jigo anyway;) Each mentioned choice would have been sportsmanlike, IMO.)

I think criticism should be directed to the tournament practice of actually treating repetition differently sometimes. Indeed it seems good idea for rules of play to discourage such practice and maybe even state that jigo and repetition are equal unless specified otherwise. OTOH in the era of fractional komi the players don't need to compare repetition to score 0, only to score win and score loss.

Another interesting thing is that Korean rules (the versions I saw so far) seem to use the same word (~mushoubu equivalent?) for both score tie and draw on repetition. I speak zero Korean, but when I checked the originals even the Korean characters were the same. This is probably the reason behind English translations referring to both as draw.

RobertJasiek wrote:We can assign the value "no result" to the game result of a "no result" game.

Why can't we just declare "no result" to be 0.5 by default? In the case that we decide to replay games that end in "no result", why can't we just use our best estimate of our winning probability? These are all well-defined values that are comparable to 1, 0.5 and 0. We could actually just throw a coin if we like too.

Even if "no result" could not be compared to 1, 0.5 and 0 it doesn't follow that we can talk about "undecidability" in the computational sense, just that the problem is not well-defined. In other words it is ambiguous, we just don't know what to do with "no result" when choosing the strategy if we can't assign a value of the right form.

So why do you talk about "undecidable strategy" instead of "ambiguous strategy"?

Maybe this discussion is irrelevant to the renaissance rules (one reason I didn't reply with questions about everything I found interesting in what was said about "undecidable strategy") but my suggestion is to just include something like the following which I kind of stole from the laws of chess and added the points for ending the game because of "repetition".

Unless announced otherwise in advance, a player who wins his game, or wins by forfeit, scores one point (1), a player who loses his game, or forfeits scores no points (0), a player who draws his game scores a half point (0.5), and if the games ends because of repetition each player scores a half point (0.5).

kvasir wrote:Why can't we just declare "no result" to be 0.5 by default?

If no result shall occur at all, I would prefer such a declaration. However, I have not written the official Japanese rules of play, which do not have such a declaration.

why can't we just use our best estimate of our winning probability?

1) A game is a competition between its players - not of the referee's arbitrary choice.

2) Earlier strategic decisions affected both creation of a no-result and the creation of the position's global environment.

Even if "no result" could not be compared to 1, 0.5 and 0

There were Japanese pro tournaments with tournament rules explicitly distinguishing no-result from jigo.

So why do you talk about "undecidable strategy" instead of "ambiguous strategy"?

Linguistic preference is immaterial.

if the games ends because of repetition each player scores a half point (0.5).

It would be possible to use such a tournament rule. In fact. some European tournaments have used such for Japanese rules.

So a bit about ko. With this approach, threefold repetition takes over many superko roles. But even dispute-only superko (Option_A below) has a bit wider effects than Chinese rules: it affects all delayable captures/disputes. Some examples are bent4 (or its inverse) + double ko seki, and moonshine life. Of these Chinese only apply superko for the last. So alternative/refined ko rules (more specific to moonshine) may also be offered as options:

• {OPTION_A} Superko for dispute phases:
  board plays that recreate a previous position since the last pass are forbidden.
• {OPTION_B} Moonshine-superko:
  board plays that recreate the position of the opponent's last pass are forbidden.
• {OPTION_C} Minimal-superko:
  board plays that recreate the start position of the phase are forbidden.
  (that is, the original disputed position)

Option_C is the simplest. It seems barely enough to capture moonshine whatever the opponent tries, but he can prolong things a bit (though he can do that anyway, even without double ko).

Option_B seems the most interesting theoretically. It might even be proposed as a universal Go rule. Basically, PASS gains a third function (besides giving up a turn and offering a stop): it now also creates/records a "protected position" for the player. (A bit like Gérard's permanently prohibited kos, and a bit like ko passes as well).

Both new options are simpler and easier than superko since only one position needs to be remembered not whole history, and neither affect the main game. I'm still pondering, there might be problems with one or both - just posting to have them looked at with more eyes. Please do try to challenge or post objections.