Chinese Rules Question (mysterious skip)

Elom · Post by **Elom** » Wed Aug 22, 2018 3:42 am

This may seem slightly silly, but I'm a little bewildered by the compensation for handicap stones under Chinese rules.

To clarify: by 'total compensation', I mean the sum of the effects of numeral compensation and tiebreakers. By 'numeral compensation', I mean any integer values added to a colour's score, the two types being 'first-move compensation' and 'stone compensation. By 'tiebreaker', I include any fractional values added to a colour's score.

(1)
(1.1)
In Japanese rules, as I have understood them, the most accurate total compensation for black having the first move is six points of numeral compensation plus benefit of the doubt when a draw takes place in the form of an extra half point.

When go is played on a board with an odd number of intersections, black could take the last move and one extra stone played onto the board. So under Chinese Rules, an additional extra point of numeral compensation is given to white for the extra point black may gain this way, changing the total compensation to seven-and-a-half-points.

What I wonder is why the last point of stone compensation given to white in even games, changing the total compensation from six-and-a-half to seven-and-a-half, is removed along with first move compensation in one-stone handicap games by giving white just half a point's total compensation instead of one-and-a-half, and then, apparently arbitrarily, added back on in games of two stones or more by giving white two-and-a-half points of total compensation.

(1.2)
In a related note: as I understand it, AGA rules consistently remove the first stone compensation point across all handicap games by employing n-1, in that n-1 points of stone compensation are added to whites score in addition to the tiebreaker, so we have half-a-point for one stone, one-and-a-half points for two stones, and so on.

From what I can tell, area scoring used in Chinese rules make it very straightforward to determine the end of a game as all positions can be played out. However, counting can be tedious, with different tricks and techniques having been developed to jet-stream the counting process. One could argue that over time, these techniques become second nature and pose little problem; nevertheless, territory scoring under Japanese rules was invented which avoided drawn out counting procedures, but this came at the cost of introducing a few mind-frazzling rules for special positions, and not without controversy. So, as most know, AGA rules combine the best of both worlds by disguising area scoring as territory scoring through the use of pass stones, so that passing costs points in the same way playing does. AGA rules being adjusting territory so that it matches the area score at it's heart, white must pass last and give an extra pass stone to lose an extra point of territory if black plays last, to correlate with black getting an extra point in area score from playing the last move in Chinese Rules. As in it's area-scoring oriental counterpart, this is recognised in AGA rules by adding a point to white's six points of first move compensation to make seven points of numeral compensation. But this point of stone compensation is removed in handicap games by employing n-1 rather than, simply, n. While we could decide to take this loss of one point as just a part of the handicap, would it not be more elegant to simply decide whether to compensate white for black's stones or not, and use either n or nothing.

(2)
(2.1)
Black plays last?
(2.2)
Couldn't black give white stones at the beginning of the game for first-move compensation and an additional stone for every move played before white's first?

edit: (3)
The half-point tiebreaker gives white victory when the integer points in a game are the same. If that is the case, would it make sense for black to have the benefit of the doubt in situations such as superko and chosei? I remember reading that AI tends to believe a seven-and-a-half point total compensation's good for white, so might this adjustment put things right?

Pio2001 · Post by **Pio2001** » Wed Aug 22, 2018 4:21 am

Hi,
As far as I understand (other will correct me if I'm wrong), the komi was increased from 6.5 to 7.5 for a different reason : it is because the win/loss ratio still favored Black with a komi of 6.5. One point was added so that the game becomes more even.

The compensation in handicap games is logical in AGA rule, because it allows the equivalence with the territory counting.

The compensation in chinese rules is strange, because it introduces a one point difference between territory counting (used mentally by chinese players) and area counting. Maybe it was based on the equivalence with the difference in komi, so that for any game, White has one more point in chinese rule than in japanese rule : because of the different komi in even games, and because of the compensation stones in handicap games.

EdLee · Post by **EdLee** » Wed Aug 22, 2018 5:12 am

Elom, sorry, I don't understand anything about komi in area scoring in handicap situations;
Just thinking out loud for the odd komi in area scoring in non-handicap, even games.

... :

Tryss · Post by **Tryss** » Wed Aug 22, 2018 11:32 am

Pio2001 wrote:Hi,
As far as I understand (other will correct me if I'm wrong), the komi was increased from 6.5 to 7.5 for a different reason : it is because the win/loss ratio still favored Black with a komi of 6.5. One point was added so that the game becomes more even.

In area scoring, the board difference in score in nearly always odd.

There is 361 points to divide in two. If there is no seki and black has n points, then white has 361-n points, and the difference between black and white score in 2n-361 : an odd number

To have an even score difference, you need a seki with an odd number of dame impossible to remove. In practice, 6.5 komi in area scoring is mostly equal to 5.5 komi

Calvin Clark · Post by **Calvin Clark** » Wed Aug 22, 2018 12:41 pm

First, I'd like to ask which text of Chinese rules you are using. I am curious myself about the origins.

Second, I agree. Komis in area scoring results should be of the form (2n + 1) + 0.5 (dividing by 2 if using half-counting, but that doesn't matter). See this SL page. (Perhaps Robert Jasiek will chime in, as he wrote that page.)

So a "no-komi" game should properly have a tie-breaker komi of 1.5 (or -0.5, but tradition is to favor white). Some Chinese players award 1 point for each handicap stone in a handicap game.

AGA rules provide a point compensation if scoring area-style, for each handicap stone after the first.

Consider these games:

Click Here To Show Diagram Code: [go]$$ Even or no komi, black played first $$ ----------- $$ | . . . . . | $$ | . . . . . | $$ | X X X X X | $$ | O O O O O | $$ | . . . . . | $$ -----------[/go]

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

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

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

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

AGA rules provide a point compensation if scoring area-style, for each handicap stone after the first. In the following tables, the score is the sum of the second to fourth columns. The score is reported from Black's point of view. The column on Pass Stones, used only under AGA territory scoring, is the net change to the store as a consequence of the pass stone rule.

AGA Rules, Territory Counting:

Code: Select all

| Handicap | T Count    | Pass Stones | -Komi | Handicap Adj. | Score |
| 0        | 5          | 0           | - 7.5 | 0             | -2.5  |
| 1        | 5          | 0           | - 0.5 | 0             | 4.5   |
| 2        | 3          | 1           | - 0.5 | 0             | 3.5   |
| 3        | 2          | 1           | - 0.5 | 0             | 2.5   |
| 4        | 1          | 1           | - 0.5 | 0             | 1.5   |
| 5        | 0          | 1           | - 0.5 | 0             | 0.5   |

AGA Rules, Area Counting

Code: Select all

| Handicap | A Count    | Pass Stones | -Komi | Handicap Adj. | Score |
| 0        | 5          | 0           | - 7.5 | 0             | -2.5  |
| 1        | 5          | 0           | - 0.5 | 0             | 4.5   |
| 2        | 5          | 0           | - 0.5 | -1            | 3.5   |
| 3        | 5          | 0           | - 0.5 | -2            | 2.5   |
| 4        | 5          | 0           | - 0.5 | -3            | 1.5   |
| 5        | 5          | 0           | - 0.5 | -4            | 0.5   |

(Note the Area and Territory Scoring are the same.)

Chinese, assuming point for each handicap stone:

Code: Select all

| Handicap | A Count    | Pass Stones | -Komi | Handicap Adj. | Score |
| 0        | 5          | 0           | - 7.5 | 0             | -2.5  |
| 1        | 5          | 0           | - 0.5 | 0             | 4.5   |
| 2        | 5          | 0           | - 0.5 | -2            | 2.5   |
| 3        | 5          | 0           | - 0.5 | -3            | 1.5   |
| 4        | 5          | 0           | - 0.5 | -4            | 0.5   |
| 5        | 5          | 0           | - 0.5 | -5            | -0.5  |

Now, that's a lot of data, but there are couple of things to note. One is that handicap scores can be almost anything. In an even game under AGA rules with 7.5 komi, if I see a result of B+0.5, I wonder if it's either a weird game with an odd number of points in seki or if the players either miscounted or didn't use pass stones. (The last is by far the most likely.) It's the wrong parity. But with handicap games, it's different as you see. The result could change.

However, the trope is that handicap games don't matter, so no one cares anyway.

Pio2001 · Post by **Pio2001** » Wed Aug 22, 2018 4:02 pm

Elom wrote:would it not be more elegant to simply decide whether to compensate white for black's stones or not, and use either n or nothing.

That would prevent the use of prisoners to count the final score.
You would have to count all the stones on the board, and rely on tables indicating the limit between victory and defeat. This limit being different according to the komi and according to the handicap (unless you want to count all the white stones and all the black stones !).

Using n-1 points allow to count the game filling territory with prisoners, dead stones and pass stones. That's much easier. And the method is the same for both handicap games and even games.

Elom · Post by **Elom** » Wed Aug 29, 2018 6:32 am

Thanks for your posts and sorry for taking a while, I'm not well versed in this and can sometimes be slow to read and assimilate quickly recently.

(I didn't explain myself well in the first post. The compensation for handicap stones only relates to area scoring; territory scoring doesn't need it!)

As stated above, odd komi in Area scoring makes sense because of the odd number of points on a 19x19 board.

Pio2001 wrote:Hi,

The compensation in handicap games is logical in AGA rule, because it allows the equivalence with the territory counting.

Because of the above, my perspective so far is:

In even games played with AGA rules and the odd 7.5 Total compensation or scoring with Area gives the same result as with Territory.
In handicap games with the odd 1.5 compensation, this is only the case when black has an odd handicap. In handicap games with the even 0.5 komi, this is only the case when black has an even handicap.

A simple solution to make Area scoring agree with the winner by territory in all normal games is to interpret first-move compensation as 6 and give white a point of compensation for every move black plays before white, which works on all odd boards such as 19x19 (361 points).

Another solution is to give white 1.5 points of compensation and two points of compensation for every two black moves before white's first, starting from move three. This gives a more accurate score but is more tricky to implement and therefore more prone to human error, hence my preference for the first.

A long example (poorly explained):

Pio2001 wrote:
Elom wrote:would it not be more elegant to simply decide whether to compensate white for black's stones or not, and use either n or nothing.
That would prevent the use of prisoners to count the final score.
You would have to count all the stones on the board, and rely on tables indicating the limit between victory and defeat. This limit being different according to the komi and according to the handicap (unless you want to count all the white stones and all the black stones !).

Using n-1 points allow to count the game filling territory with prisoners, dead stones and pass stones. That's much easier. And the method is the same for both handicap games and even games.

I think that n in area scoring agrees with territory scoring, not n-1, so I must be missing something still...

Bill Spight · Post by **Bill Spight** » Wed Aug 29, 2018 11:13 am

Elom wrote:I think that n in area scoring agrees with territory scoring, not n-1, so I must be missing something still...

From what I heard years ago (which may not be accurate), long before the AGA rules were created, the n-1 adjustment for handicap stones under Chinese rules was not to make the scoring equivalent to Japanese scoring, but was just a historical feature of the rules, long before komi came along. Maybe the feeling was that Black should not get credit for the extra stones, I dunno. Anyway, the modern Chinese rules do not have that feature, but the AGA rules do.

Pio2001 · Post by **Pio2001** » Wed Aug 29, 2018 11:57 am

The main idea is that if you use pass stones, then counting area or counting territory minus dead stones minus prisoners minus pass stones give you the same result.
This rule is true in handicap games if and only if White gets a n-1 compensation, because it stands as long as the number of white stones on the board is the same as the number of black stones on the board when you start counting.

The fact that the komi is odd, even, integer, fractional, or that the goban has an odd or an even number of intersections has absolutely no effect on the above.

The parity comes into play when we ask whether counting area or territory can change the outcome of the game.

Here is a table listing the conditions for the last pass stone under AGA rules to be able to change the winner. Or, if you prefer, for the counting style (area or territory) to be able to change the winner while the komi is 7.5.

Last play	Handicap	komi	Intersections in seki	The outcome may depend on the counting style
White				No
Black	0	7.5	0 or even	No
Black	0	7.5	Odd	Yes
Black	Even	0.5	0 ou even	No
Black	Even	0.5	Odd	Yes
Black	Odd	0.5	0 or even	Yes
Black	Odd	0.5	Odd	No

Common case : in a 9 stones game, if there are no seki, the last pass stone can change the winner.

EDIT : this is only true if there were no pass during the game.

Elom · Post by **Elom** » Thu Aug 30, 2018 7:09 am

Okay, so would this summary be an accurate way of showing things in handicap games with 0.5 komi...

If there are 0 or an even number of neutral mutual life points (the most common case):
-For even handicaps, (2, 4, 6, and 8) n-1 gives the same outcome for either counting style.
-For odd handicaps, (1, 3, 5, 7, and 9) n gives the same outcome for either counting style.

(With an odd number of neutral mutual life points, everything is reversed.
-For even handicaps, (2, 4, 6, and 8) n gives the same outcome for either counting style.
-For odd handicaps, (1, 3, 5, 7, and 9) n-1 gives the same outcome for either counting style.)

edit: as an even (0 being so) number of neutral points is the most common case, would the best ruling be that with even number of handicap stones use n-1, and with an odd number of handicap stones, use n?

RobertJasiek · Post by **RobertJasiek** » Thu Aug 30, 2018 7:18 am

1 handicap does not mean 1 handicap stone.

Elom · Post by **Elom** » Thu Aug 30, 2018 7:46 am

Calvin Clark wrote:First, I'd like to ask which text of Chinese rules you are using. I am curious myself about the origins.

I haven't looked up a chinese rules, but I did read this study:
http://www.fqjr.qc.ca/go/pdf/publicatio ... unting.pdf

Also; in light of the likelihood that white has a slight advantage, I wondered about awarding black the win when unusual situations such as triple kos occur upon the board. If a triple ko is 'bad luck', I'm sure few would mind it leaving play— white would, of course, not allow it. But chosei is meant to be a marvel... And may be a shame to lose. Perhaps limiting black's automatic win to multiple ko's is okay?

Elom · Post by **Elom** » Thu Aug 30, 2018 7:49 am

Bill Spight wrote:
Elom wrote:I think that n in area scoring agrees with territory scoring, not n-1, so I must be missing something still...
From what I heard years ago (which may not be accurate), long before the AGA rules were created, the n-1 adjustment for handicap stones under Chinese rules was not to make the scoring equivalent to Japanese scoring, but was just a historical feature of the rules, long before komi came along. Maybe the feeling was that Black should not get credit for the extra stones, I dunno. Anyway, the modern Chinese rules do not have that feature, but the AGA rules do.

Interesting! I suspect that was the case, and that AGA rules just adapted it to match AGA rules territory scoring.

Pio2001 · Post by **Pio2001** » Thu Aug 30, 2018 11:10 am

Elom wrote:Okay, so would this summary be an accurate way of showing things in handicap games with 0.5 komi...

If there are 0 or an even number of neutral mutual life points (the most common case):
-For even handicaps, (2, 4, 6, and 8) n-1 gives the same outcome for either counting style.
-For odd handicaps, (1, 3, 5, 7, and 9) n gives the same outcome for either counting style.

You mean under AGA rules ?
With or without pass stones ?

Elom · Post by **Elom** » Fri Aug 31, 2018 7:14 am

Pio2001 wrote:
Elom wrote:Okay, so would this summary be an accurate way of showing things in handicap games with 0.5 komi...

If there are 0 or an even number of neutral mutual life points (the most common case):
-For even handicaps, (2, 4, 6, and 8) n-1 gives the same outcome for either counting style.
-For odd handicaps, (1, 3, 5, 7, and 9) n gives the same outcome for either counting style.
You mean under AGA rules ?
With or without pass stones ?

AGA rules with pass stones.

Life In 19x19

Chinese Rules Question (mysterious skip)

Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)

Re: Chinese Rules Question (mysterious skip)