This 'n' that

[Sennahoj]

I have a question about one of the variations:

[Bill Spight]

I have a question about one of the variations:

does not qualify as a sente because it is a gote. It gains 5 points and then gains 4 points in reply, so that the sente sequence gains 1 point. I agree that the proverb is somewhat confusing, especially as traditionally plays are classified as double sente, which may gain something.

I hoped that the reason for the proverb that sente gains nothing would be clear from the method of multiples. I do not classify plays as double sente (except maybe for seki and such, which do not gain points). Rather double sente has to do with whole board situations, and is not intrinsic to plays or positions per se.

[John Fairbairn]

does not qualify as a sente because it is a gote. It gains 5 points and then gains 4 points in reply, so that the sente sequence gains 1 point.

If you move White 2 one point to the left it is a gote and gains 4 points. Black then does not have to reply with 4 at once. So the only dynamic difference between the two White 2s is that the correct one is - well, we have to have a name for it, so why not call it sente. Making Black reply and "gaining" an extra point needs some way of describing that attribute of the correct White 2. 'Sente sequence' certainly doesn't cut the mustard.

Seems to me you could just as well say that White 2 does not qualify as a sente because it is a reverse sente.

Conclusion: trying to explain your new approach using the traditional terms is a cure worse than the disease.

Or I've lost the plot completely, which I accept is perfectly possible.

[Sennahoj]

Get it... I think

[Bill Spight]

does not qualify as a sente because it is a gote. It gains 5 points and then gains 4 points in reply, so that the sente sequence gains 1 point.

If you move White 2 one point to the left it is a gote and gains 4 points.

Actually, it gains nothing (on average). And it allows Black to gain 1 point in reverse sente.

Here is the equivalent position from #72.

$$W Gote
$$ . . . . . . . . . . . . . . .
$$ . O O O O O O X X X X X X X .
$$ . O X X X X . . . O O O O X .
$$ . O O O O O O O X X X X X X .
$$ . . . . . . . . . . . . . . .

Click Here To Show Diagram Code

[go]$$W Gote
$$ . . . . . . . . . . . . . . .
$$ . O O O O O O X X X X X X X .
$$ . O X X X X . . . O O O O X .
$$ . O O O O O O O X X X X X X .
$$ . . . . . . . . . . . . . . .[/go]

The count in the corridor is 0, as indicated by the symmetry.

$$W Zero gain
$$ . . . . . . . . . . . . . . .
$$ . O O O O O O X X X X X X X .
$$ . O X X X X 1 3 4 O O O O X .
$$ . O O O O O O O X X X X X X .
$$ . . . . . . . . . . . . . . .

Click Here To Show Diagram Code

[go]$$W Zero gain
$$ . . . . . . . . . . . . . . .
$$ . O O O O O O X X X X X X X .
$$ . O X X X X 1 3 4 O O O O X .
$$ . O O O O O O O X X X X X X .
$$ . . . . . . . . . . . . . . .[/go]

After the count is still 0. White has a 1 point sente left at 3, which results in a local net score of 0.

'Sente sequence' certainly doesn't cut the mustard.

I'm fine with gote sequence.

I have added your to the main line of the SGF file below. It should be clear that it is no better than taking a dame. Edit: Actually, it is better on this board than taking a dame. My bad.

Conclusion: trying to explain your new approach using the traditional terms is a cure worse than the disease.

Or I've lost the plot completely, which I accept is perfectly possible.

Traditional usage of gote, sente, and double sente is something of a muddle. But it does not take much to clear things up. For instance, if you show

$$W Gote
$$ . . . . . . . . . . . . . . .
$$ . O O O O O O X X X X X X X .
$$ . O X X X X . 1 . O O O O X .
$$ . O O O O O O O X X X X X X .
$$ . . . . . . . . . . . . . . .

Click Here To Show Diagram Code

[go]$$W Gote
$$ . . . . . . . . . . . . . . .
$$ . O O O O O O X X X X X X X .
$$ . O X X X X . 1 . O O O O X .
$$ . O O O O O O O X X X X X X .
$$ . . . . . . . . . . . . . . .[/go]

to O Meien and ask him what kind of play is, he will say gote in a flash. Ask him how much it gains, and he will say 5 points.

[John Fairbairn]

For instance, if you show to O Meien and ask him what kind of play is, he will say gote in a flash. Ask him how much it gains, and he will say 5 points.

Since O Meien had to write a book specially to convey his ideas and felt compelled to give his system a new name (Absolute Counting) this is Alice in Wonderland stuff - words just mean what the Red Queen wants them to mean.

O's book is not specially well written, anyway. He belongs to the Cheshire Cat school of writing - grins at you then disappears - which seems common among mathematicians. But quoting him on this forum seems a bit perverse since there is no English version yet.

[Bill Spight]

For instance, if you show to O Meien and ask him what kind of play is, he will say gote in a flash. Ask him how much it gains, and he will say 5 points.

Since O Meien had to write a book specially to convey his ideas and felt compelled to give his system a new name (Absolute Counting) this is Alice in Wonderland stuff - words just mean what the Red Queen wants them to mean.

Well, traditionally this play is considered a gote with a swing value of 10 points. Some pros might be confused, but I feel sure that O Meien would not be. Nor would Takagawa or Shimamura or Hayashi Gembi have been, to name a few. No Humpty Dumpty or Wittgenstein here.

[Bill Spight]

I learn more about sente and gote

Back to ancient times. I originally learned about sente from the English version of Korschelt, about the value of taking sente instead of following your opponent around and of keeping sente when possible. I also learned about sente plays, which threaten the opponent with severe loss (according to Korschelt). Later, as an SDK, I learned about the classification of plays as gote, sente, reverse sente, and double sente. I also learned the rule about doubling the swing value of sente, without explanation. I also learned about assuming that sente are played when estimating territory, and the saying that sente gains nothing. (The two go together, and I understood that. If sente gained something, then it would not make sense to assume that sente are played when estimating territory, as that would change the estimate.)

As I said, I began to have doubts about double sente, in particular the double kosumi on the second line. I saw one pro game where it was played in gote, and after that I noticed that pros often left it on the board for some time without playing it, a violation of the saying to hurry to play double sente.

As I also mentioned, on my own I developed a probabilistic understanding of gote and sente, at least in evaluating positions and plays. Since we assume that sente are played, the probability of playing a sente is 100%, and the probability of playing a gote is 50%. That gave me the correct values. I did not think about it, but that did not allow for evaluating double sente, as it implied that each player would get to play the double sente.

I finally understood the rule about doubling the value of sente, when I read somewhere about evaluating a simple ko at 2/3 of its swing value. The point was how much each move gained. There is one move difference in a sente, two moves difference in a gote, and three moves difference in a simple ko. You double the swing value of sente instead of taking one half the swing value of gote because people find it easier to multiply by two than to divide by two. I did not think of the implication for double sente, dividing by zero.

I had also forgotten Korschelt’s idea of sente as threatening a severe loss. I told beginners that a play was sente if it carried a threat that was larger than anything else on the board. Not exactly accurate, but close enough.

But I was still operating by the seat of my pants. Often whether a play was sente or gote was obvious, but not always. And what about plays that were sente, but did not carry a threat that was larger than anything else on the board? Obviously they could not be played with sente yet, but presumably they would eventually, with very high probability.

Normally, as the ambient temperature dropped (as we now say), there would come a time when the largest play elsewhere would be smaller than the threat of the sente but larger than the reverse sente. At that point the sente could be played with sente. There might be exceptions, but that was the general rule. So we started with the idea that a play is sente if it carries a threat larger than the largest play elsewhere on the board, and we get a new idea that a play is sente if it carries a threat that is larger than the reverse sente. Eureka!

Later that got refined to the idea that a play is sente if it raises the local temperature and initiates a sequence of play that drops below the local temperature with a play by the opponent.

[Abyssinica]

A play is sente only if your opponent answers.

[Bill Spight]

A play is sente only if your opponent answers.

Even Korschelt did not go that far. A play is made with sente if your opponent answers.

[Bill Spight]

I thought it might be fun to use the method of multiples to evaluate one of the so called special rulings of the 1949 Japanese rules, Three Points Without Capturing. (See http://senseis.xmp.net/?TorazuSanmoku ). The ruling is attributed to Honinbo Shuwa. Shuwa did not leave a record of his reasoning, but it accords with the result of the method of multiples.



When Black plays first, White answers locally, yielding sente but making 3 points in the corner. Black starts in the next corner with the same result: 3 points for White in each corner, as advertised.

When White plays first, White gets only 2 points in the corner, but takes gote, so that Black has to start in all of the other corners. Each new corner adds 3 points for White, which is consistent with the ruling.

The fact that White gets only 2 points in gote indicates that the local temperature is -1, that is, that each play loses 1 point.

[Bill Spight]

Here is another ruling from the Japanese 1949 rules. Five Points Without Capturing. See http://senseis.xmp.net/?TorazuGomoku



Waddaya know? By the method of multiples, if Black plays first White gets 5 points in each corner, but if White plays first he gets only 3 points in each corner.

A true double sente!

[Sennahoj]

It doesn't seem like the multiples get used much in these examples

[Abyssinica]

Waddaya know? By the method of multiples, if Black plays first White gets 5 points in each corner, but if White plays first he gets only 3 points in each corner.

A true double sente!

Is this like a thousand year double sente.

[Bill Spight]

I write the rules

When I was a 3 dan I moved to New Mexico, making me the strongest player in the state. Before long I was elected president of the New Mexico Go Association, and I started holding four tournaments a year. I became aware of the problems beginners in the West have with ending the game. I was already aware of the problems they had with telling whether stones were alive or dead at the end. After about a year as president, I drafted and proposed rules that used play to decide questions of life and death. Since at the end of the game each play costs one point, whether by filling in one's own territory, or by playing inside the opponent's territory and becoming a prisoner, my idea, which I had already used with beginners, was to require that each player make the same number of moves in the play to decide life and death. That way the score would stay the same.

The rules also had a komi of 6.5 points. We were ahead of our time.

Now, I don't know how about how Honinbo Shuwa decided on the score for Three Points Without Capturing, but the New Mexico rules produced the same result. Let's see how that would work. In the top right corner we have a group with three points for Black. If the top left position is worth three points for White, then the local score in the combined positions should be zero. How does that play out?



As we already know, if Black plays first White gets three points in the top left after an even number of plays, but if White plays first he only gets two points there, but he has made one more play than Black. When Black makes the required move to make the number of plays the same for each player, the result in the two corners is zero, as advertised.

OK. What about the Five Points Without Capturing double sente? Let's take a look.



Under New Mexico rules it is worth only three points. To avoid the result of five points for White, Black to play fills a point in the right hand corner. Then White has to start play in the top left corner, and the local net score is again zero.