Please, Kirby, the gain per move is only 4 pts. Gote swing value is about the gain from two moves (one by Black, one by White), not one.Kirby wrote:Swing value can be confusing, but I think you are on the right track.
It already seems clear to you that black can get 7 points by playing first (3 captures + 1 more point territory). That’s how much black gets, from black’s perspective.
But there’s additional value in playing there: black prevented white from making points. If white had no chance to make points there anyway, blacks would gain the 7 points, and nothing else. But the fact is, if white plays there first, white gets 1 point.
So black didn’t only gain the 7 points - he gained the fact that he denied white 1 point. That’s why you add them together to get 8.
Please help with this exercise III
-
Bill Spight
- Honinbo
- Posts: 10905
- Joined: Wed Apr 21, 2010 1:24 pm
- Has thanked: 3651 times
- Been thanked: 3373 times
Re: Please help with this exercise III
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
-
Bill Spight
- Honinbo
- Posts: 10905
- Joined: Wed Apr 21, 2010 1:24 pm
- Has thanked: 3651 times
- Been thanked: 3373 times
Re: Please help with this exercise III
OK, Jika, let's get down to basics. 
How much territory does White have?
How much territory does White have?
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
- EdLee
- Honinbo
- Posts: 8859
- Joined: Sat Apr 24, 2010 6:49 pm
- GD Posts: 312
- Location: Santa Barbara, CA
- Has thanked: 349 times
- Been thanked: 2070 times
-
Kirby
- Honinbo
- Posts: 9553
- Joined: Wed Feb 24, 2010 6:04 pm
- GD Posts: 0
- KGS: Kirby
- Tygem: 커비라고해
- Has thanked: 1583 times
- Been thanked: 1707 times
Re: Please help with this exercise III
I didn’t say per move. I think it is simpler for a beginner to understand the concrete number of points he gets. By playing there, black has 7 countable points, and I think it’s simpler to think about it that way. I understand that there are technical terms here, and we can talk about per move gain, but I don’t think it helps more than confuses someone interested in the basics.Bill Spight wrote:Please, Kirby, the gain per move is only 4 pts. Gote swing value is about the gain from two moves (one by Black, one by White), not one.Kirby wrote:Swing value can be confusing, but I think you are on the right track.
It already seems clear to you that black can get 7 points by playing first (3 captures + 1 more point territory). That’s how much black gets, from black’s perspective.
But there’s additional value in playing there: black prevented white from making points. If white had no chance to make points there anyway, blacks would gain the 7 points, and nothing else. But the fact is, if white plays there first, white gets 1 point.
So black didn’t only gain the 7 points - he gained the fact that he denied white 1 point. That’s why you add them together to get 8.
I think the goal here should be to explain as simply as possible, and I’m not sure all of the posts here have that in mind.
be immersed
-
Bill Spight
- Honinbo
- Posts: 10905
- Joined: Wed Apr 21, 2010 1:24 pm
- Has thanked: 3651 times
- Been thanked: 3373 times
Re: Please help with this exercise III
I agree.Kirby wrote:I didn’t say per move. I think it is simpler for a beginner to understand the concrete number of points he gets. By playing there, black has 7 countable points, and I think it’s simpler to think about it that way. I understand that there are technical terms here, and we can talk about per move gain, but I don’t think it helps more than confuses someone interested in the basics.Bill Spight wrote:Please, Kirby, the gain per move is only 4 pts. Gote swing value is about the gain from two moves (one by Black, one by White), not one.Kirby wrote:Swing value can be confusing, but I think you are on the right track.
It already seems clear to you that black can get 7 points by playing first (3 captures + 1 more point territory). That’s how much black gets, from black’s perspective.
But there’s additional value in playing there: black prevented white from making points. If white had no chance to make points there anyway, blacks would gain the 7 points, and nothing else. But the fact is, if white plays there first, white gets 1 point.
So black didn’t only gain the 7 points - he gained the fact that he denied white 1 point. That’s why you add them together to get 8.
I agree. But simplicity does not excuse inaccuracy. There is no gain between the two positions, one worth 7 pts. to Black, and one worth 1 pt. to White, because there is no way to move from one to the other. Swing value does not indicate the gain from playing at any point. The belief that it does causes confusion, as I mentioned before. You seemed to agree.I think the goal here should be to explain as simply as possible, and I’m not sure all of the posts here have that in mind.
Using Ed's diagram:
From A Black can move to position B, and White can move to position C, both below.
7 pts. for Black in the corner.
1 pt. for White in the corner.
Black gains something from her move, and White gains something from his. Their average gain is (7 + 1)/2. That is, one half the swing value.
There is no gain for moving from B to C or vice versa, because neither player can do that. The difference between the territory values of B and C is the swing value of playing from A, and can be used to compare plays, But, as I said before, trying to use them for gains and losses leads to confusion and mistakes. I have seen players from beginners to 5 dans make those mistakes.
It is simple and accurate to say that when Black plays first she gets 7 pts., and when White plays first he gets 1 pt. There is no need to talk about gains. Especially as the confusion between swing values and gains is a known problem.
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
-
Bill Spight
- Honinbo
- Posts: 10905
- Joined: Wed Apr 21, 2010 1:24 pm
- Has thanked: 3651 times
- Been thanked: 3373 times
Re:
All you have to do is look at the four possible variations of correct play.EdLee wrote:Hi Bill,
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
- EdLee
- Honinbo
- Posts: 8859
- Joined: Sat Apr 24, 2010 6:49 pm
- GD Posts: 312
- Location: Santa Barbara, CA
- Has thanked: 349 times
- Been thanked: 2070 times
Hi Bill,
). Also, I didn't know and am still very unclear about swing value. 
Digesting.
Thanks.
That was my mistake ( or at least one of my mistakes ) in post#2: I didn't know, although I had a suspicious feeling, that the term gain has a specific meaning in existing literatures, and may be different from the everyday usage (e.g. in electrical engineeringIt is simple and accurate to say that when Black plays first she gets 7 pts., and when White plays first he gets 1 pt. There is no need to talk about gains. Especially as the confusion between swing values and gains is a known problem.
Digesting.
- EdLee
- Honinbo
- Posts: 8859
- Joined: Sat Apr 24, 2010 6:49 pm
- GD Posts: 312
- Location: Santa Barbara, CA
- Has thanked: 349 times
- Been thanked: 2070 times
Hi Bill,
Thanks for the patience with these 30k questions.
local points ) - (
local points ), correct ?
Thanks for the patience with these 30k questions.
...which is defined as (1) If the position has a territorial score, that is the territorial value.
This is the new info from post#16; I haven't looked up mean value theorem as applied to Go, yet.2) Otherwise, if the position is a non-ko position, it has a mean territorial value, defined as the average result in the alternating play when Black plays first and when White plays first in 2ⁿ copies of the position, either exactly or in the limit as n approaches infinity.
-
Kirby
- Honinbo
- Posts: 9553
- Joined: Wed Feb 24, 2010 6:04 pm
- GD Posts: 0
- KGS: Kirby
- Tygem: 커비라고해
- Has thanked: 1583 times
- Been thanked: 1707 times
Re: Please help with this exercise III
Using the term "gain" is inaccurate only if you take it on to have a particular meaning. You are using "gains" as a specific term, and I think it adds to the confusion, because it requires people to know about this terminology. Maybe "gain" has a specific meaning in some literature, but to somebody that hasn't read this literature, I think it's hard to know that it means something different than "get". In English, the two words are pretty similar. And I agree that it can be dangerous when thinking of this as a plus or minus to your score in actual play. But is it necessary to go into that detail to answer this question?Bill Spight wrote: It is simple and accurate to say that when Black plays first she gets 7 pts., and when White plays first he gets 1 pt. There is no need to talk about gains. Especially as the confusion between swing values and gains is a known problem.
Jike posted in post #28 that he was still confused. That means there were more than 25 responses that didn't make things clear to him. It means that our approach up until now, albeit precise, isn't clear to a beginner. If you are hung up about my use of the word "gain" instead of "get", I apologize, but I don't think the current discussion is easy for a beginner to digest. We should avoid adding a bunch of terminology for folks to learn, and keep things simple here. We can have a separate thread that's not in the beginner section that goes into all of these terms and nuances.
Local position value, territorial score, ko situations, thermography, fuzzy logic, and possibilities have all been discussed in this beginner thread. Is it really necessary to answer the basic question?
The only reason I am posting in this thread is in an attempt to simplify things - I don't think the current >30 posts are very clear to a beginner, however technically accurate they may be.
---
So I'll try again.
If black plays first in that diagram, black GETS 7 points. If white plays first, white GETS 1 point. It's useful to consider both possibilities in swing values, because by playing first, black can GET 7 points and prevent white from GETTING 1 point. We take both of these benefits into account, and that's why they're summed together.
If white could not GET any points by playing first, but black could still GET 7 points by playing first, then black playing there would still be useful - but not as useful as this scenario, since it doesn't prevent white from GETTING 1 point like it does here.
In contrast, if white could GET, say 2 points by playing first, and black can GET 7 points by playing first, then such a play would be more valuable than this one - even though black can still GET 7 points, it prevents white from GETTING 2 points, which is more than what white would GET if he plays to get the 1 point in this position.
I hope this is more clear.
be immersed
-
Bill Spight
- Honinbo
- Posts: 10905
- Joined: Wed Apr 21, 2010 1:24 pm
- Has thanked: 3651 times
- Been thanked: 3373 times
Re:
Ed, I am using gain in its ordinary English meaning. KISS (Keep It Simple, Sister) is my motto.EdLee wrote:Hi Bill,That was my mistake ( or at least one of my mistakes ) in post#2: I didn't know, although I had a suspicious feeling, that the term gain has a specific meaning in existing literatures, and may be different from the everyday usage (e.g. in electrical engineeringIt is simple and accurate to say that when Black plays first she gets 7 pts., and when White plays first he gets 1 pt. There is no need to talk about gains. Especially as the confusion between swing values and gains is a known problem.). Also, I didn't know and am still very unclear about swing value.
Digesting.Thanks.
But here and elsewhere people want formality, so I oblige.
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
-
Bill Spight
- Honinbo
- Posts: 10905
- Joined: Wed Apr 21, 2010 1:24 pm
- Has thanked: 3651 times
- Been thanked: 3373 times
Re: Please help with this exercise III
Actually, I am using gain in its simple, straightforward English meaning. I am doing so to dispel the confusion caused when people have been taught swing values. The confusion is that people use swing values when they should use gains.Kirby wrote:Using the term "gain" is inaccurate only if you take it on to have a particular meaning. You are using "gains" as a specific term, and I think it adds to the confusion, because it requires people to know about this terminology.Bill Spight wrote: It is simple and accurate to say that when Black plays first she gets 7 pts., and when White plays first he gets 1 pt. There is no need to talk about gains. Especially as the confusion between swing values and gains is a known problem.
Simple example: Suppose that with komi I am behind by 1½ pts. and get the last play, which has a swing value of 2 pts. I think that I will win by ½ pt., but, to my surprise, I lose by ½ pt. The reason is that the play only gained 1 pt. Simple, straightforward English.
Now, the errors that people make in real life because they confuse swing values with gains are usually more complicated, but the principle is the same. Gains and losses add and subtract; swing values do not, at least as people try to add and subtract them.
Ever since getting on rec.games.go around 25 years ago I have tried to dispel this confusion. To use gain to explain swing values, as you and Ed both have done, without noting that for gote there are two gains involved, one gain made by Black when she moves first, and one gain made by White when he moves first, makes it difficult for a beginner to avoid confusing swing values with gains, it seems to me. That is why I have asked you both not to do so.
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
- jlt
- Gosei
- Posts: 1786
- Joined: Wed Dec 14, 2016 3:59 am
- GD Posts: 0
- Has thanked: 185 times
- Been thanked: 495 times
Re: Please help with this exercise III
@Jika: my reasoning is very down-to-earth. I just look at the triangled area
If Black moves first then Black captures 3 stones and gets 7 points in that area, so in that area, the score (Black's points minus White's points) is 7.
If White moves first then White gets 1 point, so the score (Black's points minus White's points) is -1.
The rest of the board doesn't change, so the difference between these two situations is 7-(-1) = 8.
If Black moves first then Black captures 3 stones and gets 7 points in that area, so in that area, the score (Black's points minus White's points) is 7.
If White moves first then White gets 1 point, so the score (Black's points minus White's points) is -1.
The rest of the board doesn't change, so the difference between these two situations is 7-(-1) = 8.
-
lightvector
- Lives in sente
- Posts: 759
- Joined: Sat Jun 19, 2010 10:11 pm
- Rank: maybe 2d
- GD Posts: 0
- Has thanked: 114 times
- Been thanked: 916 times
Re: Please help with this exercise III
I sort of agree with Bill. Saying a move "gains" X points suggests a few things to me, via the normal English language meaning of the word:
(#1) The difference in how much a position is "worth" before and after the move is X points.
(#2) If I were to pass instead, I should get a final result X points worse on average, because obviously a pass "gains" 0 points instead of X points.
It's not an absolute requirement of course, but it would still be nice to have the word "gain" mean something consistent with the above two intuitions. And in double-gote positions in real games, both of these intuitive meanings are satisfied by the half the swing value:
In situation "a" black has 0 points, in "c" black has 1 point. In situation "b", in real game situations like this, on average either player is equally likely to get it first. So on average black has 1/2 point.
If black does spend a move in situation "b", black will turn it into position "c". Black's move gains half of a point because before they had 1/2 point on average, afterward they have 1 point. The difference in how much these positions is worth is half a point. (matching intuitive meaning #1 above)
If black were to pass rather than playing, on average black would do 1/2 point worse. For example, if there were an odd number of positions like this left and nothing else, black end up a full 1 point worse. But if there were an even number of them, black would not end up worse at all for passing. In a real game, it's basically random whether there will be an even or odd number, so on average, black will be 1/2 point worse off by passing. (matching intuitive meaning #2 above)
Similarly, if you took random pro games with a position like "b" on the board (where neither black nor white had any liberty or eye issues, both white and black were already 100% alive), and let white get a free stone at "b" near the start of the endgame, not spending their turn, and then had both players finish the endgame normally and count, on average black would do 1/2 point worse, not 1 point worse. (another variation on intuitive meaning #2 above)
Explaining this requires precision on the part of the teacher, but I think if the teacher is precise, it's not particularly hard for the beginner to understand gains rather than swing values. When I teach beginners in-person, I do actually make sure to be precise when I explain move values, and the people I've taught seem to understand pretty well.
-------------------------------------------------------------------------------------------------------
So back to the position at hand:
With this position, if I was going to very wordily explain it in real life to someone, I'd say something like:
"If black plays first, black has 7 points in the corner. If white plays first, white has 1 point, which is the same as (-1) for black. If neither player has played here yet and is busy with fighting elsewhere on the board, we expect on average 3 points for black here since knowing nothing else, either player might get to move first. So relative to that, the value/gain/worth of a move here is 4 points. If black plays first, black will do 4 points better than that (7 points) and if white plays first, white will do 4 points better than that (-1 points). Either way, the value of a move here is 4 points."
And also in practice, if I was continuing to try to explain, I'd also say something like:
"Many people you talk to might call this 8 points, because they're counting the total swing between black playing first and white playing first as 8 points. Note that the 8 points is actually over two moves, yours and your opponents. For example, if it was your turn, and you passed up playing here to play somewhere else instead, and then your opponent then played here instead of you, you'd get to play somewhere else again, - you would be a total of 8 points worse here compared to playing here yourself, but you'd have gotten two moves elsewhere instead, not just one, so the per-move gain/loss is still 4.
Per-move values work better once you get into more complex positions that don't get finished in one move or two moves, or you start getting into cases where moves might be sente or forcing. But lots of people will talk in terms of total swing. If you personally also find it easier to think in terms of total swings for the simplest cases like this, that's also okay, just be aware of the difference and make sure you're consistent".
(#1) The difference in how much a position is "worth" before and after the move is X points.
(#2) If I were to pass instead, I should get a final result X points worse on average, because obviously a pass "gains" 0 points instead of X points.
It's not an absolute requirement of course, but it would still be nice to have the word "gain" mean something consistent with the above two intuitions. And in double-gote positions in real games, both of these intuitive meanings are satisfied by the half the swing value:
In situation "a" black has 0 points, in "c" black has 1 point. In situation "b", in real game situations like this, on average either player is equally likely to get it first. So on average black has 1/2 point.
If black does spend a move in situation "b", black will turn it into position "c". Black's move gains half of a point because before they had 1/2 point on average, afterward they have 1 point. The difference in how much these positions is worth is half a point. (matching intuitive meaning #1 above)
If black were to pass rather than playing, on average black would do 1/2 point worse. For example, if there were an odd number of positions like this left and nothing else, black end up a full 1 point worse. But if there were an even number of them, black would not end up worse at all for passing. In a real game, it's basically random whether there will be an even or odd number, so on average, black will be 1/2 point worse off by passing. (matching intuitive meaning #2 above)
Similarly, if you took random pro games with a position like "b" on the board (where neither black nor white had any liberty or eye issues, both white and black were already 100% alive), and let white get a free stone at "b" near the start of the endgame, not spending their turn, and then had both players finish the endgame normally and count, on average black would do 1/2 point worse, not 1 point worse. (another variation on intuitive meaning #2 above)
Explaining this requires precision on the part of the teacher, but I think if the teacher is precise, it's not particularly hard for the beginner to understand gains rather than swing values. When I teach beginners in-person, I do actually make sure to be precise when I explain move values, and the people I've taught seem to understand pretty well.
-------------------------------------------------------------------------------------------------------
So back to the position at hand:
With this position, if I was going to very wordily explain it in real life to someone, I'd say something like:
"If black plays first, black has 7 points in the corner. If white plays first, white has 1 point, which is the same as (-1) for black. If neither player has played here yet and is busy with fighting elsewhere on the board, we expect on average 3 points for black here since knowing nothing else, either player might get to move first. So relative to that, the value/gain/worth of a move here is 4 points. If black plays first, black will do 4 points better than that (7 points) and if white plays first, white will do 4 points better than that (-1 points). Either way, the value of a move here is 4 points."
And also in practice, if I was continuing to try to explain, I'd also say something like:
"Many people you talk to might call this 8 points, because they're counting the total swing between black playing first and white playing first as 8 points. Note that the 8 points is actually over two moves, yours and your opponents. For example, if it was your turn, and you passed up playing here to play somewhere else instead, and then your opponent then played here instead of you, you'd get to play somewhere else again, - you would be a total of 8 points worse here compared to playing here yourself, but you'd have gotten two moves elsewhere instead, not just one, so the per-move gain/loss is still 4.
Per-move values work better once you get into more complex positions that don't get finished in one move or two moves, or you start getting into cases where moves might be sente or forcing. But lots of people will talk in terms of total swing. If you personally also find it easier to think in terms of total swings for the simplest cases like this, that's also okay, just be aware of the difference and make sure you're consistent".
-
Bill Spight
- Honinbo
- Posts: 10905
- Joined: Wed Apr 21, 2010 1:24 pm
- Has thanked: 3651 times
- Been thanked: 3373 times
Re: Please help with this exercise III
Many thanks to lightvector for taking the time and effort to explain things so well.
Obviously, precision in endgame evaluation is not very important. I can easily construct a problem where a mistake of 1/64 pt. or less will lose the game, but how often will such a mistake matter in real life? Still, IMHO an amateur shodan should be able to play the small endgame almost perfectly. And endgame concepts can apply at any stage of the game.
I first learned endgame evaluation from the books of Sakata and Takagawa. Takagawa did a good job of evaluating plays, but both of them taught the evaluation of positions at the same time. I still remember all those Xs in Sakata's diagrams indicating points that would become territory if one player played first but would not become territory if the other player did.
From Jika's remarks I get the impression that 321go did not start with the evaluation of positions, or teach it together with the evaluation of plays. But if anything, the evaluation of positions should come first. The evaluation of positions provides the conceptual basis for understanding the evaluation of plays. That is why I posted the basic problem that I did.
Obviously, precision in endgame evaluation is not very important. I can easily construct a problem where a mistake of 1/64 pt. or less will lose the game, but how often will such a mistake matter in real life? Still, IMHO an amateur shodan should be able to play the small endgame almost perfectly. And endgame concepts can apply at any stage of the game.
I first learned endgame evaluation from the books of Sakata and Takagawa. Takagawa did a good job of evaluating plays, but both of them taught the evaluation of positions at the same time. I still remember all those Xs in Sakata's diagrams indicating points that would become territory if one player played first but would not become territory if the other player did.
From Jika's remarks I get the impression that 321go did not start with the evaluation of positions, or teach it together with the evaluation of plays. But if anything, the evaluation of positions should come first. The evaluation of positions provides the conceptual basis for understanding the evaluation of plays. That is why I posted the basic problem that I did.
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
At some point, doesn't thinking have to go on?
— Winona Adkins
Visualize whirled peas.
Everything with love. Stay safe.
-
Jika
- Lives with ko
- Posts: 162
- Joined: Sat May 25, 2019 12:09 am
- GD Posts: 0
- Has thanked: 145 times
- Been thanked: 17 times
Re: Please help with this exercise III
Okay, a lot to read!
First, I like Kirby's approach to explain things so simply that I might understand.
I get a bit confused by the more advanced discussions, but I don't really mind being confused.
I'm only worried I fail expectations and waste your time.
But, as I posted earlier, if I get the impression that my stupid questions spark discussions that are interesting to you all, it's good too. I can skim the discussion and if necessary ask again.
You might decide this is cluttering the beginners' section.
So, either one could split such a thread, or maybe I post in a different place?
Would a study journal be a better place?
But please know I'm happy and grateful for all answers, whether I understand them or not (in study journal I might revisit them later; only, it must be hard to keep track of all the information one gets).
First, I like Kirby's approach to explain things so simply that I might understand.
I get a bit confused by the more advanced discussions, but I don't really mind being confused.
I'm only worried I fail expectations and waste your time.
But, as I posted earlier, if I get the impression that my stupid questions spark discussions that are interesting to you all, it's good too. I can skim the discussion and if necessary ask again.
You might decide this is cluttering the beginners' section.
So, either one could split such a thread, or maybe I post in a different place?
Would a study journal be a better place?
But please know I'm happy and grateful for all answers, whether I understand them or not (in study journal I might revisit them later; only, it must be hard to keep track of all the information one gets).