Stating this Ruleset as Compactly as Possible

For discussing go rule sets and rule theory
User avatar
HermanHiddema
Gosei
Posts: 2011
Joined: Tue Apr 20, 2010 10:08 am
Rank: Dutch 4D
GD Posts: 645
Universal go server handle: herminator
Location: Groningen, NL
Has thanked: 202 times
Been thanked: 1086 times

Re: Stating this Ruleset as Compactly as Possible

Post by HermanHiddema »

Mr. Mormon wrote:I'm not stubborn. What's a better approach that's as short?


Don't optimize prematurely. All this worrying about length of words and dropping punctuation is irrelevant until you have confidence your rules say what you want them to say. You can always change "intersections" into "grid points" or "points" into "spots" later.

First try several different approaches at stating the rules to find what works (e.g. chains vs. paths vs. grid colorings vs. whatever).

Don't worry about grammatical legality, worry about readability. The purpose of language is to communicate, not to satisfy some set of mathematical rules about punctuation, spelling, and word order. (e.g. my approach with "own" is much more understandable than your "One's", regardless of grammatical legality.)

Don't include needless details. IMO the "etc." allowing for more than just Black and White is useless.

Quitting is a meta-game rule, not a game rule, as is the whole agree/resume stage of counting. These are not required for complete game rules. If you want it in there, just add some "Players may agree on the result, or" clause before the counting rules.
Mr. Mormon
Dies with sente
Posts: 99
Joined: Tue Aug 24, 2010 3:44 am
GD Posts: 0
KGS: MrMormon
Has thanked: 11 times
Been thanked: 2 times

Re: Stating this Ruleset as Compactly as Possible

Post by Mr. Mormon »

HermanHiddema wrote:Don't worry about grammatical legality, worry about readability. The purpose of language is to communicate
This is a theoretical problem, a challenge.
HermanHiddema wrote:First try several different approaches at stating the rules to find what works (e.g. chains vs. paths vs. grid colorings vs. whatever).
Fair. I originally thought about the approaches and concluded paths was the most concise, but I typed the others out this time and:

Path:

Captured is without paths along the lines to empty spots through only One's.
Area is One's and empty spots without paths to Other's not through One's.

Recursive:

Not captured is 1 grid line from empty spots or the not captured.
Area is One's and empty spots 1 line from area.

Chain:

2 connected pieces are 1 grid line apart or connected to a third.
Not captured is connected to a piece 1 grid line from an empty spot.
Area is One's and empty spots not connected to an empty spot 1 line from Other's.

There's probably a way to improve the chain definition, but recursion seems to be better than paths after all. So here's a CMNZR attempt:

1. Black, White etc. quits, passes, or puts One's piece on an empty grid point, removing Other's captured (not 1 grid line from empty spots or Other's uncaptured) then One's captured-avoid board states of One's past turns.
2. If One passes following a pass Other resumes 1. by no pass or agrees on pieces to remove before each finds area (+7 if White if Black had no first free turns). Area is One's pieces and empty spots 1 line from area.

Compare with CMNZR 1.1:

1. Black, White etc. quits, passes, or puts One's piece on an empty grid point, removing Other's without paths along the lines to empty spots through only Other's then for One's-avoid board states of One's past turns.
2. If One passes following a pass Other resumes 1 by no pass or agrees on pieces to remove before each sums One's, empty spots without paths to Other's not through One's, and 7 if White if Black had no first free turns.

Surprisingly (to me too), CMNZR 1.1 is three characters shorter. I think what's wrong with the recursive version is the necessary introduction of the term "captured" to enable recursion, and the change of focus from pieces to captured pieces in rule 1 means that "pieces" can't be omitted from "One's pieces" in the area definition.

Now back to some other comments you made. There is an implied "each" after "Black, White etc."; with the right emphasis, it sounds fine to me in speech. Proper nouns can be used to capitalize common nouns, like Enemy or Justice. Removing "etc." would both remove another of the few indicators of alternating turns and leave the grammatically incorrect "Black, White". Fixing it with "Black then White" makes the ruleset actually contradict itself by implying just two turns.

We seem to disagree on the definition of "complete game rules". Here's my take. The game is Go. Rules specify what are legal actions by a nonempty set of intelligent entities while voluntarily constrained to those rules. Complete means all-encompassing of actions that must be allowed and disallowed to constitute a game. I have up to now willed that everything common to Go rules (except for whether winning has degrees - I'm not sure yet) be included in CMNZR. Quitting is part of the game; it is ending the game. How to handle a (dead) stone disagreement, namely by resuming a phase with a specific player, is certainly a rigorous and precise action like how situational superko works. Cheating without being caught, not quitting, is where I draw the line of meta-game - that's where an entity demonstrates a lack of will to be constrained by the rules.

Someone may convince me that how to decide who goes first should also be included by my reasoning despite not being included in most Go rulesets...
Mr. Mormon
Dies with sente
Posts: 99
Joined: Tue Aug 24, 2010 3:44 am
GD Posts: 0
KGS: MrMormon
Has thanked: 11 times
Been thanked: 2 times

Re: Stating this Ruleset as Compactly as Possible

Post by Mr. Mormon »

I am starting to realize some optimizations for the recursive approach, though. Here's CMNZR 2.0:

1. Black, White etc. quits, passes, or puts One's piece on an empty grid point; remove Other's captured (not 1 grid line from empty spots or Other's uncaptured) then One's-avoid board states of One's past turns.
2. If One passes following a pass Other resumes 1. by no pass or agrees on pieces to remove before each finds area (empty spots 1 line from area or One's pieces; +7 if White if Black had no first free turns).

"captured" could maybe be replaced by "caught", "doomed" (lol), "dead", or "out".
RobertJasiek
Judan
Posts: 6272
Joined: Tue Apr 27, 2010 8:54 pm
GD Posts: 0
Been thanked: 797 times
Contact:

Re: Stating this Ruleset as Compactly as Possible

Post by RobertJasiek »

Mr. Mormon wrote:How many characters


Why characters? Surely you should first define a formal meta-language and then express the rules in it. Like

US

stands for "unique situations", i.e., it is required to create only not repeated situations. And

TM

stands for "an allowed alternate turn order sequence pattern is player-mixed", where player-mixed means Black - White. Etc. Of course, you want to omit blanks, so we get

USTM

Needless to say, further compression is possible. The entire rules can be expressed in a single character, which points to the rules description:

0

I do not think though that zero characters suffice, as palapiku seems to imply.
Mr. Mormon
Dies with sente
Posts: 99
Joined: Tue Aug 24, 2010 3:44 am
GD Posts: 0
KGS: MrMormon
Has thanked: 11 times
Been thanked: 2 times

Re: Stating this Ruleset as Compactly as Possible

Post by Mr. Mormon »

I don't see how defining new terms (which will take characters in English) will provide optimizations, as there is very little redundancy of terms in the rules right now. I could abbreviate Black and White, but that's probably about it. And yeah...pointing defeats the purpose, as I started this thread with a much shorter description than CMNZR that references NZ rules.
Post Reply