Bonobo wrote:
W L7, B connects at L6, W L8?
That gains 0.5 point at K-08. Only 0.5 point because we already count 0.5 point for White there.
White has a slightly bigger play in the same area.
| Life In 19x19 http://www.lifein19x19.com/ |
|
| This 'n' that http://www.lifein19x19.com/viewtopic.php?f=12&t=12327 |
Page 7 of 53 |
| Author: | Bonobo [ Thu Nov 26, 2015 8:00 am ] |
| Post subject: | Re: This 'n' that |
W L7, B connects at L6, W L8? |
|
| Author: | Bill Spight [ Thu Nov 26, 2015 8:47 am ] |
| Post subject: | Re: This 'n' that |
Bonobo wrote: W L7, B connects at L6, W L8? That gains 0.5 point at K-08. Only 0.5 point because we already count 0.5 point for White there. White has a slightly bigger play in the same area.
|
|
| Author: | Bill Spight [ Thu Nov 26, 2015 9:29 am ] |
| Post subject: | Re: This 'n' that |
Endgame continuation:  and  are ambiguous plays.  is a reverse sente. Each gains 1 point.I daresay that Black had read the endgame out at this point and earlier, but he could have gotten the last play at temperature 1. is the last play that gains 1 point. Black can get away with it because the threat at 5 is bigger than the threat at 6. Size matters.  Now, because there are smaller plays left, getting the last one point play does not matter. That is because the final scores at go are integers. David Wolfe, one of the authors of Mathemacial Go, pointed out to me that, with no kos the player to start the play below temperature 1 can "round up" the fractional count to the next integer in his favor. I may add that, since a 1/3 point ko is with very rare exceptions the smallest play before the dame, it will normally be the last play of the game with territory scoring. Thus, we may reasonably expect that, with Black getting the last 1 point play here, White filling the ko to save the  stone will be the last play to round the score to the nearest integer. (In this case, it is not hard to read the game out, so we can be sure.)In the actual game Black started play below temperature 1, so we cannot say that the ko will not be fought. Without reading the possible ko fight out, there is the possibility that Black will lose the ko fight and not be able to round up the score in his favor. If he had gotten the last one point play he would not have that concern. (But as I said, surely Black had already read the game out. )Two questions: What is correct play to finish the game after the actual game position in the first diagram? What is correct play to finish the game after the position in the second diagram? OC, there may be more than one answer for each question. Happy Thanksgiving! 
|
|
| Author: | Bill Spight [ Sun Nov 29, 2015 8:49 pm ] |
| Post subject: | Re: This 'n' that |
I hope everyone had a good Thanksgiving, or a good weekend if you do not celebrate Thanksgiving. Here are the final plays of the game. Black won the final ko, for jigo. The Gokyo Seimyo, by Hayashi Genbi, which was published in 1835, has examples of swing counting, and I expect that the players had accurate values for the plays at temperature 1. Intatsu surely knew that  -  was worth a bit less than that, since Black's space is open below . (It gains 5/6 of a point, as does  .)However,  gains only 3/4 of a point, while  gains 5/6. This slight misstep by Black means that he had to win the final ko to get jigo. (OC, he had surely read the ko fight out. ) (or ) gains 5/6.  gains 3/4. gains 2/3. - gains 1/2.  gains 1/3. Each player fills one ko. Result: jigo. Edit: Added the other variation below. Since Black got the last play at temperature 1, - gains 5/6,  gains 3/4,  gains 1/3. Result: jigo. Again, without Black having to win the final ko.
|
|
| Author: | Bill Spight [ Tue Dec 01, 2015 12:43 pm ] |
| Post subject: | Re: This 'n' that |
Earlier, Schachus opined that the swing value of - below was worth a bit more than 7 points. Later, Dave (ez4u) showed the value of the ko formed by -  , but even without the ko there is a sense in which Schachus is right. is not sente (it is ambiguous), but is reverse sente to preserve the point of White territory at "a". has preserved the point of Black territory at "b". The local play is over. Even without playing the ko, - gains 1 point. Considering only the points "a" and "b" to be at stake, the local count after White has captured the Black stone on B-02 is 0. After Black's follow-up with sente the local score is 0; after White follow-up it is -1.A lot of people have trouble with the saying that sente gains nothing, because it certainly seems to gain something by eliminating the reverse sente. But, as we have seen, playing a sente (or ambiguous play) with sente does not gain any points. However, when the play made with sente is final, as here (and the sente side has no gainful gote options), CGT tells us that the sente has gained something. The sente has gained something, but that something is not a number. If you try to assign a number to the difference, it does not work. But White is better off before the Black follow-up, even if you cannot measure how much better off numerically.
|
|
| Author: | Bill Spight [ Tue Dec 01, 2015 3:26 pm ] |
| Post subject: | Re: This 'n' that |
When to play reverse sente? Let us revisit the diagram below. After  should White play the reverse sente at "a"? After all, it is the hottest play on the board, as it gains 3 points. "b" would gain only 2.5 points.As we know, the answer is no. Way back when, I came up with the rule that White should play the reverse sente if 3 > 5 - 2 + 1 = 4 3 being the value of the reverse sente and the numbers on the right side being the swing values of the gote, in descending order. The answer is no, mainly because of the large drop in temperature between the largest gote and the second largest gote. Edit: BTW, the play in question does not have to be global reverse sente now, its threat only has to be larger than the second largest gote in the environment. Unlike the case with the global sente, approximation is not a special problem. Below are the successive approximations, as we look more deeply into the board. 3 > 2.5 --> Play reverse sente. 3 < 5 - 1 = 4 --> Play top gote. 3 < 5 - 2 + 0.5 = 3.5 -- Play top gote. These successive approximations illustrate the significance of the temperature drop. Such a drop, while not all that unusual, is not the norm. In normal environments the hottest play is very likely to be the best play. Way back when, I used to construct problems with environments consisting of simple gote. They were not only unrealistic, but also a bit dull.  Now I use normal environments in most of my studies and problems. That makes them more realistic, and, I hope, more interesting and more convincing. The downside is the possibility of making oversights, such as the one that Dave found with the ko. 
|
|
| Author: | Bill Spight [ Tue Dec 01, 2015 4:18 pm ] |
| Post subject: | Re: This 'n' that |
What is a ko threat? When I was learning go, the textbooks gave the following answer: A ko threat is a move played in order to win the ko. The threat must be larger than the value of the ko. As we would now say, the threat must be hotter than the ko. The books said that the swing value of the threat must be larger than 2/3 the difference between winning and losing the (regular) ko. When I began my ko research, I quickly found that the textbooks were wrong on both counts. Winning the ko is only one purpose of playing a ko threat, and the size of a regular ko has little or nothing to do with the size of a ko threat. Generally we compare the size of a ko threat with the temperature of the environment, not with the size of the ko. Some threats can be smaller. Also, I identified three reasons for playing a ko threat: 1) To win the ko (whether successful or not); 2) To gain something in exchange for losing the ko; 3) To minimize the loss when the ko winner has excess threats. As for the form of a ko threat, there is no single answer. As a rule, ko threats are sente, but there are cases where the largest gote may be considered a ko threat. Some people regard any board play as a ko threat, because it lifts the ban on retaking the ko. There is something to that. However, if every play is a ko threat, then no play is a ko threat. The term loses its meaning. More when I get into my ko research.
|
|
| Author: | Bill Spight [ Thu Dec 03, 2015 1:27 pm ] |
| Post subject: | Re: This 'n' that |
Thermography (i) Thermography is a graphical technique for finding the mean value and temperature of combinatorial games. When applied to a non-ko go position it finds its territorial count and miai value. The application of thermography to ko and superko positions is not obvious, as they are not combinatorial games. Professor Berlekamp discovered a way to do so for single ko positions using the concept of komaster. A komaster can win a ko but cannot reduce the temperature of the environment. For the first time, complex kos such as 10,000 year kos and approach kos could be evaluated. Their values depend upon who is komaster. Now, my approach to ko evaluation was rather different. I included an environment, but the standard assumption was that the temperature of the environment (although I did not use that term, yet) was typically dropping. I took as my standard of value the largest gote in the environment, and evaluated the ko along with the rest of the environment, including other gote and ko threats. I called that collection the ko ensemble. (Obviously I was not getting the standard values of kos. Which I considered an improvement. ) One place my approach foundered was with approach kos and 10,000 year kos. Not that I did not get some interesting results, but in general the analysis was too complex to be practical or tractable. Such kos and their ko fights were not easy to conceptualize. The idea of komaster allowed Professor Berlekamp to understand such complex kos much better.However, the idea of komaster did not apply to multiple kos. After struggling for a couple of years with the evaluation of double kos, I realized how I could reconcile thermography with an environment with temperature drops. By redefining thermography in terms of play in an ideal environment, I could apply it to multiple kos, while getting the same results for the rest of go positions. (Usually when you extend a definition in math, you have to give up something, but not in this case. )
|
|
| Author: | Bill Spight [ Sat Dec 05, 2015 1:09 am ] |
| Post subject: | Re: This 'n' that |
Seeing vs. calculation of variations As readers here may have noticed, this is a theme of mine. Both seeing and calculation of variations are part of reading. My sense is that in general the calculation of variations is given more weight, if seeing is considered at all. Perhaps that is because seeing is somewhat mysterious, as it involves unconscious mental processing. I am a champion of seeing, at least for amateurs. Pros can fend for themselves. I do not mean to disparage the calculation of variations, but I do not believe that it is the royal road to improvement. I first ran across the term, seeing, in a bridge book by Victor Mollo, that I bought for my sister, who was a home player. Seeing is probably more important in bridge than in go, since bridge is a game of hidden information. Mollo opined that amateur players should focus less on tactical calculations and more on seeing. IMO, that advice applies to kyu players at go. Along those lines, I recall a remark by a Korean 5 dan who was visiting the Los Alamos, New Mexico, club: "Why do they take so long, when they have nothing to think about?" Here is an example of seeing in a problem posed here recently at viewtopic.php?f=15&t=12490 . It is a subproblem of one of the problems from the Gokyo Shumyo. First, you can see that this atari does not work. Black does not have the room to make two eyes. Besides, Black can make an eye by capturing the stone.A beginner may not see that sequence, but it is familiar to experienced players. As is White's killing move.Along with , takes away the potential eye at  . Experienced players know that is safe. Now, you may consider - to be the calculation of a variation, but it is something that experienced players see quickly.The next two diagrams illustrate my use of seeing to solve this subproblem. I have shown the corner after because that was the image in my mind at this point. However, I did not treat the corner as an eye, but asked how Black could make life in the shaded region. That question led me to see the next diagram. The stones produce two eyes on the points. Seeing that led to the following calculation. prevents the cramped life in the previous diagram, but then makes the second eye in the corner. The calculation in the next diagram is unnecessary, as the result has already been seen. :Now, a lot of seeing is simply familiarity. And, OC, this problem is solvable by calculation alone. However, instead of relying upon calculation alone, I invoked seeing to tell me what stones might be necessary for life. Edit: Oh, yes. A relaxed state of mind aids seeing.
|
|
| Author: | Bill Spight [ Sat Dec 05, 2015 1:03 pm ] |
| Post subject: | Re: This 'n' that |
Thermography (ii) Let’s draw a couple of thermographs. We know that in this position in an environment of simple gote that each gain t points, the result after Black captures the stone and White replies in the environment is 1 - t, and the result after White saves the stone and Black replies in the environment is -1 + t. We found that it is indifferent who plays first when t = 1 and the value of the position, v, is thus 0, by solving the equationsv = 1 - t v = -1 + t We can find the solution graphically by drawing lines with those equations. They intersect at (v,t) = (0,1). Attachment: gote01.png [ 2.57 KiB | Viewed 20782 times ] Note that CGT reverses the convention for the horizontal axis, so that numbers to the left are greater than numbers to the right. Now, below temperature 1, this graph is easy to interpret. Each player prefers to make the local play than to play in the environment, with a result of 1 - t if Black plays first and -1 + t if White plays first. But above temperature 1 each player prefers to play in the environment. There is no local result per se, but as we expect the value of 0 when the temperature drops to 1, we take that as the value. Attachment: gote01z.png [ 2.41 KiB | Viewed 20782 times ] We indicate that by drawing a vertical mast at 0 above temperature 1. |
|
| Author: | Bill Spight [ Mon Dec 07, 2015 9:36 am ] |
| Post subject: | Re: This 'n' that |
Thermography (iii) Here is another easy position to calculate. Let’s draw its thermograph. elsewhere saves the stone and captures the stone for a local score of -3. Black plays in the environment, gaining t points. That gives us the line, v = -3 + t. captures and threatens to save . captures . The net local score is 0. That gives us the vertical line, v = 0. (Note that a vertical line indicates a line of play with the same number of plays per side. Here it is one play each, for the mast it is 0 plays each.) But this does not look like a sente. What do we do in this case? elsewhereWhat line does that give us? Well, it is not hard to calculate the local count. It is 2 for the stone minus 1 below, or 1. That gives us the line, v = 1 - t. But this is supposed to be a graphical method. Let’s draw the thermograph for this position.Attachment: gote01b.png [ 2.43 KiB | Viewed 20782 times ] Now let’s shift that thermograph one move to the right, by subtracting t from it. Attachment: scaffold 02.png [ 2.5 KiB | Viewed 20782 times ] Now the right wall of the shifted thermograph gives us the lines, v = 0 and v = 1 - t. (We are no longer interested in the left wall of the thermograph.) v = 0 is relevant when t ≤ 1 and v = 1 - t is relevant when t ≥ 1. (We can write the shifted right wall as v = min(0, 1-t). The thermograph indicates minimax play at ambient temperature, t.) Now we are ready to draw the thermograph of the original position. Attachment: gote002.png [ 3.07 KiB | Viewed 20782 times ] The thermograph tells us that the local count of the original position is -1, and that it is a gote where each play gains 2 points. Not only that, it tells us that when the ambient temperature is less than 1, Black can capture the White stone with sente.
|
|
| Author: | Bill Spight [ Mon Dec 07, 2015 9:26 pm ] |
| Post subject: | Re: This 'n' that |
Thermography (iv) Let’s do another one. Here is another easy position to calculate. elsewhereThis looks familiar. Without going through all the steps, we know that to form the right side of our thermograph we will make use of the lines, v = -7 and v = -8 + t, combined as v = max(-7, -8+t). Like this: Attachment: rt scaffold 03.png [ 2.65 KiB | Viewed 20779 times ] BTW, we call this the right scaffold of the thermograph. elsewhereWell, this does not look like gote, but we can tell which lines we use to construct the left scaffold of the thermograph. v = -5 and v = -1 - t. The equation for the scaffold is v = min(-5, -1-t). Like this: Attachment: lt scaffold 03.png [ 3.23 KiB | Viewed 20779 times ] We put them together to construct the thermograph. Attachment: sente 03.png [ 3.66 KiB | Viewed 20779 times ] The walls of the thermograph are colored. The right wall is colored red up to temperature 3, and the left wall is colored blue up to temperature 4. Above that the mast is colored black. The colors show who plays first in the indicated line of play. Blue indicates that Black plays first, red indicates that White plays first, black indicates that neither player plays, and purple indicates that either player plays first. The thermograph tells us that the local count of the original position is -5, and that it is a sente where the reverse sente gains 3 points. It also tells us that when the ambient temperature is between 3 and 4, Black (normally) has the privilege of capturing the White stone with sente. Also, when the temperature is below 1, this is a double sente. Edit: Corrected "gote" to "sente". Oops!
|
|
| Author: | Bill Spight [ Thu Dec 17, 2015 11:40 am ] |
| Post subject: | Re: This 'n' that |
Sorry for the hiatus. I have some more about thermography, but I thought that a change of pace would be good. Here is something about ancient games and books. When I was learning go, a go magazine carried a review of this ancient game. Black has just played  . Where did White play?
|
|
| Author: | Bill Spight [ Fri Dec 18, 2015 12:29 pm ] |
| Post subject: | Re: This 'n' that |
OK. Here is the game continuation (hidden). IIRC, the magazine commentator was Miyamoto Naoki, 9 dan. In the format he made comments and a couple of amateurs asked questions. In discussing  Miyamoto said that Gennan Inseki was much stronger than he was. He meant stronger at fighting. Go strategy had certainly advanced since Gennan's time. At the time I discounted Miyamoto's comment, chalking it up to Oriental reverence for masters of the past. And I only got interested in ancient games in the 1990s. Now that I have come to appreciate them better, I think that Miyamoto was right. I got interested in ancient games when I stumbled across a go web site around 20 years ago which featured the site owner's extensive collection of ancient go books. He pictured them on the site, along with pictures of the pages of many of them. The site included a collection of castle games. I don't think that he showed all the game records that he had -- it's pretty tedious work to put them up --, but there were a lot of them, and I played over a number of them every month. He did not seem to be updating the site, and it went down after a year or two. Later I ran across a site that organized games by players, and had games for nearly all of the heads of the four go houses. It has just recently gone down. Among the ancients, Gennan is one of my favorites, as are Dosaku, Jowa, and Hayashi Gembi. I know of no current site devoted to ancient go, but GoGoD is a treasure trove. I have also discovered ancient books in the online libraries of the Japanese National Diet and Waseda University. It is interesting to see what has changed and what is the same. Here is the game between Gennan and Shusaku, courtesy of GoGoD. |
|
| Author: | skydyr [ Fri Dec 18, 2015 1:04 pm ] |
| Post subject: | Re: This 'n' that |
That's really interesting. I looked at both and  separately and rejected each, but did not see them in conjunction with each other.
|
|
| Author: | Bill Spight [ Sat Dec 19, 2015 1:22 am ] |
| Post subject: | Re: This 'n' that |
Old in Go I like the Mini-Chinese even more than the Chinese configuration, and it seems to have derived from the latter. In modern times, I think that that is so, but the Mini-Chinese is by far the older idea. This example occurred in 1684.  OC, it is from a handicap game, since playing the first move in the corner on the 4-4 point in even games was not revived until the late 19th century. A few years ago I ran across the Mini-Chinese in an old text as a joseki, which meant that it was played in actual games. at 7 was possible. In those days the corner approach was by no means preferred to playing on the side.Let's see how the game developed. White was Honinbo Dosaku Meijin.  invades, and play through  is familiar today. I admit that when I first started studying ancient games, a play like  took me by surprise. Not that I thought it was bad, but because my impression was that play back then was very territory oriented. So it was, by comparison with today's play, but players still had a keen appreciation of influence and the center. Dosaku, in particular, performed magic there. Dosaku was already Meijin at the time, back when 9 dan meant something. Professional ranks were more widely spaced then, so that a 9 dan would give four stones to a 2 dan. Dosaku was so strong that he was said to be 13 dan, two stones stronger than meijin.  is ideal.  could have extended farther on the bottom side, but Hoshiai was content to allow  to build a framework, and then to keep sente and attack on the left side. Black won by 12 points. BTW, I recommend ancient games to kyu players who focus on territory and have trouble with influence and thickness. Ancient players were generally territory oriented but had a good understanding of influence and thickness, too. There is a lot to learn there.
|
|
| Author: | Bill Spight [ Sun Dec 20, 2015 4:16 pm ] |
| Post subject: | Re: This 'n' that |
Old in go (ii) Another Hoshiai Hasseki game.  -  is no longer considered joseki. White's corner is too big. However, it was still considered joseki well into the 19th century, if not later. It appears without comment in Gokyo Seimyo, 1835.When I was learning go, it was considered a mistake to extend to  without making the enclosure in the top left corner first. In the 17th century players did not feel so constrained.  prevents White from making the high enclosure, which would be good for White.After a repeat in the top left corner,  makes territory and threatens to invade the top side. It also threatens to make a nice framework on the right side, with a double wing. Still, IMO it is too slow.Where do you guess White played next? This is not a problem. I can't say that it was the best play, but I like it.
|
|
| Author: | skydyr [ Mon Dec 21, 2015 7:34 am ] |
| Post subject: | Re: This 'n' that |
| Author: | Bill Spight [ Mon Dec 21, 2015 1:52 pm ] |
| Post subject: | Re: This 'n' that |
Old fashioned corner approach strikes us as unusual today, but this was an ancient joseki. Even today, the approach to the - enclosure is standard.Truth to say, would not have occurred to me, but it threatens to make a large corner territory, thus inviting .Both of these sequences appear without comment in Gokyo Seimyo. The following sequence appears in a book by Inoue Yasunobu, 5 dan, Igo GenmyoOchiboShu (1909). and attack White from the top side, after which White plays the normal looking . Now what?Inoue says that and are interesting, a form of praise in Japanese commentary. Then  leads to  and the following sequence.Then and  prevent a ladder and  serves to capture the stones.Inoue states the obvious: Very bad for White. An amusing result.
|
|
| Author: | bayu [ Mon Dec 21, 2015 2:17 pm ] |
| Post subject: | Re: This 'n' that |
Bill Spight wrote: Old in go (ii) After a repeat in the top left corner, makes territory and threatens to invade the top side. It also threatens to make a nice framework on the right side, with a double wing. Still, IMO it is too slow.Where do you guess White played next? This is not a problem. I can't say that it was the best play, but I like it. I've got a honest question about pratique. When you say: this is no joseki anymore. How do you know? I'm not disputing whether the sequence in the game above is still joseki or not. I believe you that it is outdated. I'm wondering, how could I possibly have found out. Did a book or someone who knows tell you? Do you simply observe, that it doesn't show up anymore in pro games? Kogo still has the joseki in (and says GOOD VARIATION), it appears in the Go joseki app for Android (which doesn't seem to be based on Kogo) and I was very happy when I finally learnt it. What are the sources that tell you, that something is outdated? |
|
| Page 7 of 53 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |
|