English translation of Shibano's Fuseki Revolution
-
gowan
- Gosei
- Posts: 1628
- Joined: Thu Apr 29, 2010 4:40 am
- Rank: senior player
- GD Posts: 1000
- Has thanked: 546 times
- Been thanked: 450 times
English translation of Shibano's Fuseki Revolution
Kiseido has just issued a translation by John Power of Shibano Toramaru's book Fuseki Kakumei (Fuseki Revolution). The book discusses why Chinese fuseki and Sanrensei are not so popular now, how AI play has changed attitudes toward conventional wisdom andevaluations, and new AI generated joseki.
-
CDavis7M
- Lives in sente
- Posts: 716
- Joined: Tue Jun 15, 2010 2:18 pm
- Rank: Shokyu
- GD Posts: 0
- Universal go server handle: CDavis7M
- Has thanked: 109 times
- Been thanked: 140 times
- Contact:
Re: English translation of Shibano's Fuseki Revolution
Did you get it? What do you think?gowan wrote:Kiseido has just issued a translation by John Power of Shibano Toramaru's book Fuseki Kakumei (Fuseki Revolution). The book discusses why Chinese fuseki and Sanrensei are not so popular now, how AI play has changed attitudes toward conventional wisdom andevaluations, and new AI generated joseki.
There was some more discussion on it here: viewtopic.php?f=17&t=18349
I ended up getting the book in Japanese a few months ago before the English version was announced. I've only gone though a few of the fusekis, mostly the first section. Translating is an effort so I just read the parts I was interested in. I thought about just buying it again in English but I have other books to read too.
-
gowan
- Gosei
- Posts: 1628
- Joined: Thu Apr 29, 2010 4:40 am
- Rank: senior player
- GD Posts: 1000
- Has thanked: 546 times
- Been thanked: 450 times
Re: English translation of Shibano's Fuseki Revolution
It is a useful book, readable for players at kyu level through dan levels. I couldn't say whether it is out of date already since I myself am out of date LOL (or is that expression itself out of date). I came up as a go player in the 1970's, thoroughly imbued with the Japanese approach. In fact I still like that way of playing go and I think it offers valuable insights even now that people are disparaging Japanese go. Also I side with John Fairbairn concerning time limits. I believe we need to understand what we are doing. It is no good for humans to imitate machines; just as the development of "deep learning" techniques found a way to play that suits the machine, humans should find a human way to play. I liked hearing Michael Redmond reject an AI move because he just didn't like it, and Shibano also says that he dislikes some of the AI moves. There have been "revolutions" before. I recall how different the playing of Chinese and Korean players seemed when we began to be aware of Chinese go via the visits of Chinese players to Japan and the China-Japan Supergo matches occurred, and Go World covered games of the top Korean players. Well, I got off the topic of Shibano's book, but what I find the most interesting thing about it is that Shibano explains why many of the strange-looking AI moves work.
-
hzamir
- Dies in gote
- Posts: 50
- Joined: Fri Oct 01, 2021 9:27 am
- GD Posts: 0
- Has thanked: 4 times
- Been thanked: 9 times
Re: English translation of Shibano's Fuseki Revolution
I think AI will not have really arrived until they have cracked the real AI challenge: AIs should be able to provide human comments on their own games, explaining to us humans why they work.gowan wrote:... Well, I got off the topic of Shibano's book, but what I find the most interesting thing about it is that Shibano explains why many of the strange-looking AI moves work.
-
CDavis7M
- Lives in sente
- Posts: 716
- Joined: Tue Jun 15, 2010 2:18 pm
- Rank: Shokyu
- GD Posts: 0
- Universal go server handle: CDavis7M
- Has thanked: 109 times
- Been thanked: 140 times
- Contact:
Re: English translation of Shibano's Fuseki Revolution
I agree to an extent. But I also think that it depends on how close to the AI's skill a player is, or how much stronger a player is compared to the opponent. Top pros can copy AIs because they can back it up. But for me, I'm not going to be able to successfully invade a moyo like an AI, find tesujis like an AI, and I can't be convinced that setting up an elaborate ko is a better move. So I find little point in copying AI beyond what a pro can explain to me. I like Shibano's explanations before they are short. I just did a quick check and many of the variation are only a few moves, with some being around 10 and hardly any approaching 20.gowan wrote:It is no good for humans to imitate machines; just as the development of "deep learning" techniques found a way to play that suits the machine, humans should find a human way to play. I liked hearing Michael Redmond reject an AI move because he just didn't like it, and Shibano also says that he dislikes some of the AI moves.
--------------------
I believe even before Alpha Go I was working on a technique to explain gradient boosting tree machine learning(not than neural networks). A quick search shows that Google has been working on this issue: https://cloud.google.com/explainable-ai. I don't know who would sponsor the effort though.hzamir wrote:I think AI will not have really arrived until they have cracked the real AI challenge: AIs should be able to provide human comments on their own games, explaining to us humans why they work.
-
John Fairbairn
- Oza
- Posts: 3724
- Joined: Wed Apr 21, 2010 3:09 am
- Has thanked: 20 times
- Been thanked: 4672 times
Re: English translation of Shibano's Fuseki Revolution
I read quite a bit of the original Shibano. I had to because I was writing The First Teenage Meijin.
It appealed to me because he had next to no recourse to numbers or win rates. Indeed, it wasn't even clear that he was using AI at all. He said his main approach to the topic was to look at internet games by other pros and see what they were doing differently from normal pro play, on the assumption that they had been copying AI.
This means he has been looking at the topic in away quite different from most people. At least, as far as I can judge, almost all amateurs but also many pros have been taking the approach of monkey see, monkey do. Shibano's approach has been see monkey do, what human think of it?
That's so refreshing. But it's no guarantee that he came up with useful answers. In one sense he did. He won the Meijin. But how much of that was because of psychological effects? How much was random chance?
I personally didn't find much of what he said about AI moves themselves at all useful. The greatest value for me was in what he said about traditional moves seen through the prism of AI moves. Since, as other people have remarked above, we humans still need a rationale we can handle in a practical way to play go, any insights on traditional go which is likewise based on that perspective are likely to help.
In short, just as Shibano let other pros do the donkey work, we amateurs should also let others do the monkey work.
It appealed to me because he had next to no recourse to numbers or win rates. Indeed, it wasn't even clear that he was using AI at all. He said his main approach to the topic was to look at internet games by other pros and see what they were doing differently from normal pro play, on the assumption that they had been copying AI.
This means he has been looking at the topic in away quite different from most people. At least, as far as I can judge, almost all amateurs but also many pros have been taking the approach of monkey see, monkey do. Shibano's approach has been see monkey do, what human think of it?
That's so refreshing. But it's no guarantee that he came up with useful answers. In one sense he did. He won the Meijin. But how much of that was because of psychological effects? How much was random chance?
I personally didn't find much of what he said about AI moves themselves at all useful. The greatest value for me was in what he said about traditional moves seen through the prism of AI moves. Since, as other people have remarked above, we humans still need a rationale we can handle in a practical way to play go, any insights on traditional go which is likewise based on that perspective are likely to help.
In short, just as Shibano let other pros do the donkey work, we amateurs should also let others do the monkey work.
-
Knotwilg
- Oza
- Posts: 2432
- Joined: Fri Jan 14, 2011 6:53 am
- Rank: KGS 2d OGS 1d Fox 4d
- GD Posts: 0
- KGS: Artevelde
- OGS: Knotwilg
- Online playing schedule: UTC 18:00 - 22:00
- Location: Ghent, Belgium
- Has thanked: 360 times
- Been thanked: 1021 times
- Contact:
Re: English translation of Shibano's Fuseki Revolution
It's a side note of course but I disagree. Intelligence can show through articulation but it is not restricted by it. Human experts who are inept at explaining why they know what they know, do exist, in spite of the Feynman principle. Ramanujan is a good example. He "saw" properties of numbers without being able to explain to a(nother) human how he was able to do that. It's very similar to AI Go bots showing us sequences without an explanation. In Ramanujan's case, we had good reason to doubt our own intelligence, not his. In AI's case, why should we do otherwise? We already call it artificial to distinguish the nature of the intelligence.hzamir wrote:I think AI will not have really arrived until they have cracked the real AI challenge: AIs should be able to provide human comments on their own games, explaining to us humans why they work.gowan wrote:... Well, I got off the topic of Shibano's book, but what I find the most interesting thing about it is that Shibano explains why many of the strange-looking AI moves work.
-
gowan
- Gosei
- Posts: 1628
- Joined: Thu Apr 29, 2010 4:40 am
- Rank: senior player
- GD Posts: 1000
- Has thanked: 546 times
- Been thanked: 450 times
Re: English translation of Shibano's Fuseki Revolution
When I wrote "imitate" above I meant that the human player makes an AI move only for the reason that an AI made that move in some position. In the past players were chastised for following "joseki" moves without understanding the meaning.
As for Ramanujan's mathematics, he was fortunate that Hardy was willing to spend time and hard work to validate many of Ramanujan's formulas. Ramanujan had been dismissed by other professional mathematicians and I think he might never have been recognized had Hardy not seen Ramanujan's merit. The success of AI players demonstrates the validity of the way AI's play. I wonder whether there could be a method of play that is superior to the current AI method. For example if quantum computers ever develop to a high level of practical function, maybe current AI-style go might be superseded.
As for Ramanujan's mathematics, he was fortunate that Hardy was willing to spend time and hard work to validate many of Ramanujan's formulas. Ramanujan had been dismissed by other professional mathematicians and I think he might never have been recognized had Hardy not seen Ramanujan's merit. The success of AI players demonstrates the validity of the way AI's play. I wonder whether there could be a method of play that is superior to the current AI method. For example if quantum computers ever develop to a high level of practical function, maybe current AI-style go might be superseded.
-
RobertJasiek
- Judan
- Posts: 6272
- Joined: Tue Apr 27, 2010 8:54 pm
- GD Posts: 0
- Been thanked: 797 times
- Contact:
Re: English translation of Shibano's Fuseki Revolution
My attempt to buy directly from Bozulich has failed this time because I got no clear confirmation that it would be ok.lichigo wrote:Where is it possible to buy it in Europe ?
I looked at Schaak en Gowinkel het Paard but their postage to Germany is too high for a single book.
I asked Hebsacker Verlag (in Germany) and they expect it in November. However, unless you live in Germany, their international postage is also too high, I think.
I have not noticed other European retailers listing that book (yet). It has not been at Amazon.DE or .UK.
So unless you live in the Netherlands or Germany, asking Bozulich directly might still be worth trying because airmail postage from Asia is surprisingly low.
As long as USPS is having trouble with Corona, ordering from Kiseido.USA is not really an option for Europeans.
-
CDavis7M
- Lives in sente
- Posts: 716
- Joined: Tue Jun 15, 2010 2:18 pm
- Rank: Shokyu
- GD Posts: 0
- Universal go server handle: CDavis7M
- Has thanked: 109 times
- Been thanked: 140 times
- Contact:
Re: English translation of Shibano's Fuseki Revolution
Maybe you'd be more interested in Shibano's new series 定石・布石の常識破壊 (now everything is looking like a yojijukugo idiom) called something like Destruction of Common Sense of Joseki and Fuseki with topics like:
-
RobertJasiek
- Judan
- Posts: 6272
- Joined: Tue Apr 27, 2010 8:54 pm
- GD Posts: 0
- Been thanked: 797 times
- Contact:
Re: English translation of Shibano's Fuseki Revolution
What are the reasons for your decision not to buy the book?lichigo wrote:I took a look inside but I didn't like it.
-
John Fairbairn
- Oza
- Posts: 3724
- Joined: Wed Apr 21, 2010 3:09 am
- Has thanked: 20 times
- Been thanked: 4672 times
Re: English translation of Shibano's Fuseki Revolution
Robert:
I made a few notes when I first looked at the series (I assume the book is a reprint of this). I discontinued when I lost interest, but I'm willing to share what I wrote, on the off-chance it may make some sense to you.
I made a few notes when I first looked at the series (I assume the book is a reprint of this). I discontinued when I lost interest, but I'm willing to share what I wrote, on the off-chance it may make some sense to you.
I know I shouldn't pre-judge, but I'd be rather surprised if you liked it (no numbers, etc). Many others here would appreciate it, but I suspect it would be more those who like to keep in touch with what is going on rather than those striving to get a higher rank.FUSEKI REVOLUTION
Solving the puzzle of the disappearance of old patterns
Go Weekly series by Shibano Toramaru
1. NEW SERIES ON DISAPPEARING FUSEKIS
Trains of thought brought on by AI
(2019-01-28)
First the overall picture.
Once common sanrensei and Chinese fuseki have almost disappeared over past couple of years in pro go.
But he emphasises he is not saying they are bad. It is rather that the numbers show they favour White, probably because White has found new countermeasures.
But Black could find new resources, too, and the difference is in any case only about 1 point of komi.
It is therefore no surprise that that there are pros who say: give me 2 points of komi and I'll play sanrensei and the Chinese.
Amateurs should go ahead and play them if they like them.
The openings surveyed here are:
* Sanrensei
* Chinese (including the high (Q9) and also the shorter one (R8)
* The mini-Chinese
* Star and facing kogeimajimari
* The Kobayashi
* The New Kobayashi
* Nirensei (especially the reason for the disappearance of the slide (D2 instead of K4))
(Notes: -tati plural of fuseki and use of rare de iru form.)
2. BIG MOYOS HARD TO SETTLE
Sanrensei - popular with amateurs
(2019-02-04)
Settled pro evaluation is that it is hard for Black to win: too hard to turn moyo into territory.
Moyos have gone down further in estimation since AI.