Hello everyone.

I am coming back to Go after a long hiatus. Since I discovered the game when I was 16, I have taken two breaks, each of which was pretty long. My first account, named Mithrandir, was on Pandanet. I actually spent most of this period kibitzing. I don't remember exactly what rank I reached, but it must have been close to 7k. The account seems to be have been washed away by the tide of time. In this period, I did no tsumego (didn't even know they existed). I mainly played games and kibitzed. I watched some of Dwyrin's early lectures on youtube.

After many years of occasionally watching a Go related youtube video, I made a concerted effort to get back into playing Go back in 2019 on OGS. That account I named Finrod Felagund, to fit the theme of the previous name. The experience was strange in that I had almost no reading skill but still had some deep-seated intuition. I won games just playing the spot that felt right. I attribute this mainly to watching way too many games (it was seriously an addiction at one point) and trying to guess the next move. That only took me so far, and I started to seriously do tsumego for the first time. After a few months, I got to 2k. Unfortunately, life intervened again, and I'm only now getting enough time and energy to pay attention to Go.

I'm hoping that keeping this study journal will give me a good place to keep track of my progress and to get some advice on what to do next.

My current daily Go regimen:

a.) 1-2 Serious games, reviewed with the help of Lizzie.

b.) The 6 daily Tsumego provided on the Tsumego Pro app.

-4 out of 6 of these are pretty easy for me, and I have solved some of the harder ones before. Not ideal, but very convenient.

c.) 1 video from this series (Road to be Dan Player) by Yoonwong: https://www.youtube.com/watch?v=L5-joMu2tdA

I pause the videos and try to read all variations, including the wrong paths. These have also been relatively easy, but I do not always see all the interesting ways to fail by making a ko. I also anticipate that they will get harder over time.

d.) A few problems from Gokyo Shumyo on Sensei's Library. I try not to look at the solutions until I am sure, but I have been mistaken twice so far (i.e. I thought I knew the solution for sure but had misread.)

e.) Various video lectures on youtube as time allows. This is most of the joseki study that I get.

Things I'd like advice on:

1.) I almost never count while I play, and my endgame amounts to "play Sentes and then the biggest looking Gote." Actually, most of my games are not close. I often start all or nothing fights and win or lose by a large margin. This is obviously an issue, since I am probably making overplays to compensate for my uncertain evaluation of the game and my lack of endgame skill.

I don't really know where to start on learning about yose. I have tried to find a good book on the subject, but there doesn't seem to be a widely agreed upon book. If anyone could recommend a book (or books) on counting and evaluating endgame moves, that'd be great. I'm looking for something practical but I'm not afraid of math.

2.) Should I use an AI to review my games? And if so, which one? I have some version of LeelaZero at the moment. Any advice on how to effectively review my games would be welcome.

Finally, I've attached a game of mine against a 1d. I am probably not as strong now as I was last year, but I hope to get back to that level pretty quickly.

Thanks to anyone who read this.

_________________
Now men awoke and listened to Felagund as he harped and sang, and each thought that he was in some fair dream, until he saw that his fellows were awake also beside him; but they did not speak or stir while Felagund still played

Your feat of reaching 2k through good intuition had me amazed. I'm struggling at 10k with good tsumego solving skills but never had the chance put my reading to use in real games, guess my poor shape intuition is what stopping me from improving. You mentioned about guessing moves and watch a lot of games.

Did you look at pro games and tried to guess them? If yes, which particular player's games you looked at most?

I should clarify. By the time I was 2k, my reading had improved significantly. So it's not like I was playing at 2k level without any reading.

I actually mostly watched high dan (5d-9d) games on IGS since they were a good deal faster and more convenient for me. Pros don't play perfectly, and so of course high dans make many mistakes, big and small. I think it was actually most instructive when I thought they had made a mistake. I would wrack my brain and the position until either I was sure they had made a misread or had realized that their move was right or had some deeper purpose. It helps that I would ask for explanations of stronger players who were also watching. I think for ddks, high dan play can be considered near perfect, just like as sdks or low dans, pro play can be considered near perfect.

_________________
Now men awoke and listened to Felagund as he harped and sang, and each thought that he was in some fair dream, until he saw that his fellows were awake also beside him; but they did not speak or stir while Felagund still played

I have some practical advice here: https://lifein19x19.com/viewtopic.php?p=218115#p218115

There is also a good bit of material here and on SL. As I said, probably the best place to start is with endgame tesuji. To calculate endgame positions and plays, you need to know best local play. It is true that you can find out which plays are best by using endgame evaluation techniques, but if you don't have a clue as to best play, that can be tedious.

As for evaluation, start with evaluation of positions, not plays. The evaluation of plays follows from the evaluation of positions. Caution: The traditional way of counting the size of plays and of classifying plays is confusing. I can't say that the classifications and counting method are wrong, but they are misapplied. There are sente positions and plays, gote positions and plays, and positions and plays that are ambiguous between sente and gote. See discussions here and on SL. As for the size of a gote or reverse sente play (or of a certain kind of sequence of plays) it is the difference between the value of the original position and the value of the resulting position.

As for books, I can recommend Robert Jasiek's books on the endgame.



Any good AI bot should be fine. I am partial to KataGo myself. One advantage of using a bot to review is that you can play around with different plays and see how it evaluates them. Caution: When a bot evaluates a position, it typically produces evaluations of a number of alternatives, along with the number of rollouts (playouts, visits) that it used for each. Beware of evaluation with relatively few rollouts. They can be quite inaccurate. If you want to compare those plays with others, make them yourself, so that the bot gives them a lot of rollouts. Their evaluations can change dramatically. E. g., a bot does not like your play and gives it a small number of rollouts. Make the play yourself. Sometimes the bot will change its mind. And always it will give you a good reply to your play if it is a mistake.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

About reviewing with bots:

Extreme interactivity is key. If you just look passively at the moves or variations they suggest you'll notice sequences and shapes that don't make sense to you or are surprising. So "interrogate" the bot for why.

E.g. it says your move is bad and you should tenuki and attack something else, but you don't see why that move works at all, and it even has the opponent "concede" that the move works and give up stones. Okay, so play down that line, but instead of having the opponent concede, play the move you'd expect the opponent to respond with and see how the bot responds.

E.g. it says you should play move X to threaten some stones but that the opponent should confusingly tenuki? Go ahead and have the opponent respond and see why - maybe it shows you that the opponent's saving the stones doesn't work, or works but becomes too heavy.

E.g. it says you should tenuki when you defended in game, and it has the opponent "agree" by also playing elsewhere, but you don't see why as the shape just sits there unsettled. Tenuki as it says and then have the opponent play the move you were afraid of. Maybe the bot shows you that you can just tenuki again it's because the stones aren't big enough. Maybe it shows you that actually your shape was lighter than you thought and the opponent's threat isn't a big deal. Maybe it shows you tactically that the move you were afraid of simply doesn't work.


Interactively "asking the bot questions" like this is waaaaay better than staring at the numbers or even looking at the PV that the bot gives you on any of the moves in your game.

Edit: Also a big one is to pay attention to who has sente. I've often seen some kyu players try to analyze their moves with a bot and be super-confused as to why the bot says they should prefer a given result when it's locally worse than the result they got in the game, not realizing that they have sente in this variation and gote in the other.

For endgame counting and tesuji I highly recommend Lee Chang Ho's Endgame Techniques (2 volumes), available electronically on gobooks.com. There are some other good modern endgame books that I will let others recommend, but I really like his methodical but readable approach.

On endgame:



Thanks for the recommendation. Neither those books nor that website had been on my radar.



Honestly, I have been interested in those books, but it seems like I would have to buy 4 books to get his whole collection on the endgame, and I am wary that the more theoretical of the volumes will impractical for my purposes. Once I get in the habit of regularly counting during my games, I may take the plunge on the whole set.

As for endgame tesujis, I am mostly ignorant. I've picked up the monkey jump and the always pernicious issues with liberty shortages, both from bitter experience. I suppose I will move on to the Gokyo Seimyo after the Gokyo Shumyo.





This is very good advice. My method so far has been to basically just click through the game with Lizzie, categorizing moves as good or bad depending on the AI's reaction.

Thanks to everyone who chimed in.

Journal Updates:

I haven't been able to play this weekend, but I have made some Tsumego progress.

Gokyo Shumyo: Got up to problem 67 in section 1.

Problems to come back to:

Section 1: 13, 16, 23, 38, 45, 52, 55, 56

_________________
Now men awoke and listened to Felagund as he harped and sang, and each thought that he was in some fair dream, until he saw that his fellows were awake also beside him; but they did not speak or stir while Felagund still played

Alright, I am updating my study plan slightly. Maybe if my results are good, others can use this thread for ideas.

First, I am adding some books that I read (or half-read) a long time ago. First, I have heard that rereading books can be useful. Second, I want to solidify my conceptual understanding of the game. Much of my decision making is intuitive, which is of course an important level on which to work, but it makes it difficult to review my play and improve it afterward. "Why did I play there?" It looked nice, felt right, couldn't see an obvious way for my opponent to punish. Not always so vague, but too often. So, time to study some relatively conceptual books.

The books for this stage of study:

Attack and Defense
Opening Theory Made Easy
And some books on the endgame, still being finalized.

In other news:

I ran into this problem in Gokyo Shumyo: https://senseis.xmp.net/?GokyoShumyoSection1Problem70

I didn't get this the first time I read it out, although the next day in the shower I realized my mistake. But, this problem, which is a fundamental sort of shape, revealed a gap in my knowledge, if not my reading. So, I have resolved to essentially memorize Cho Chikun's "All About Life and Death." This will serve as both some extra reading practice and improve my knowledge base.

How will I accomplish this memorization?

1.) Each day, I will read out two of the problems/shapes. Afterward, I will read Cho Chikun's commentary, which usually contains every conceivable variation. Next, I will add the shape and the variations to an sgf file.

2.) I will read out the two problems from the previous day, checking the already complete sgf files.

3.) I will finally read out two random problems from the set [1, n-2] where n is the current day after the beginning of this process.

After the whole set as been gone through, I will continue the random problems until the whole thing has been burnt into my mind. Will this lead to much improvement? Hard to say. I am hoping that if I know this information, then I will be able to speed up my judgement of Life and Death situations.

Random side note: I find it impossible to read seriously in games again katago, even with a high handicap. I guess since I know I will lose, I can't summon the energy to try to win. This is probably a deep defect in my personality and could explain some of the disappointments I have experienced in the past. Didn't Lee Sedol say that playing alphago was like looking in a mirror?

_________________
Now men awoke and listened to Felagund as he harped and sang, and each thought that he was in some fair dream, until he saw that his fellows were awake also beside him; but they did not speak or stir while Felagund still played

I am afraid that the bots have made Opening Theory Made Easy obsolete. Best to check your moves with a bot. Meanwhile, I have written up some ideas in this series: https://lifein19x19.com/viewtopic.php?f=15&t=16909



Try playing it on the 9x9 with a handicap that gives you around a 50% chance of winning. You might also set a high reverse komi for the 19x19, with the same aim.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

This is one of the things that fascinates me about go: that such different styles can work equally well at amateur level. No reading, all intuition, versus no intuition, all reading -- they both seem to top out at around amateur shodan. I've been told that Koreans use the term "flower go" for the all-intuition style.

The book 200 Endgame Problems, also available on gobooks.com , is pretty good for our level. Many of the problems appear simple at first but have hidden depths.

That thread looks interesting. I am not sure, however, that an obsolete theory is necessarily useless, or even unworthy of study. There are many cases, particularly in science, where obsolete theories (like the Bohr model) are still taught because they prepare the student for the more complete theories. In most cases, essential elements of the obsolete (or incomplete) theory is true, and the theory retains usefulness.

I don't know if Traditional Opening Theory is wrong in the same way that the Bohr model was wrong (obsolete but useful), or if it's wrong in the way that, say, the Ptolemaic model of the movement of heavenly bodies was wrong (obsolete and useless).

Maybe in any case, it would be instructive to check older books like Opening Theory made easy with the help of the bots. My main issue when learning with bots is that it's easy to find the right move, but hard to find the concept behind that move. The machine can't check my interpretation of the move, at least not directly, so there's a high chance that a weak player like myself will create an empire of dreams, taking the wrong lessons from the right moves. The advice earlier, to interrogate the AI, will certainly help with that. I don't think even pros can avoid completely the danger of misinterpretation.




Mhmmm, good idea.

[quote="xela"]

I agree. Go is awesome in part because of how it reveals aspects of human nature, learning, etc. Intuition, though, does have its limits. I have big issues with certain types of tesuji problems, because I am almost blind to strange moves that wouldn't normally be played. My own personal neural network might be too restrictive, pruning too many candidate moves before I even consider them.

_________________
Now men awoke and listened to Felagund as he harped and sang, and each thought that he was in some fair dream, until he saw that his fellows were awake also beside him; but they did not speak or stir while Felagund still played

For the most part, the moves recommended in books on the opening go back at least to the 19th century, in practice if not in theory. Many of them the bots do not like. In addition, plays that were taboo they do like. That presents a real challenge to our way of thinking about the opening, which we have not yet met. We may have to revise or even ditch old concepts and develop new ones.

One thing that seems apparent is that the bots value the corners much more than the sides. This is not exactly the old adage, corners, sides, center, because sometimes they seem to value the center more than the sides, as well. Let me give an example of corners over sides.

Click Here To Show Diagram Code

[go]$$Bc Ideal extension
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . 1 . . . . . O . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . X . . . . . . . . . . . . . O . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |[/go]

is an ideal extension from the Black corner, and would be an ideal point for White, as well. Traditional books usually show the next diagram to emphasize that point.

Click Here To Show Diagram Code

[go]$$Bc Ideal development
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . . 1 . 2 . . . O . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . X . . . . . . 3 . 4 . . . . O . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |[/go]

Not that - are made immediately, but this sequence is taken to indicate a normal development for each player, and makes it obvious that Black is glad to have gotten there first. They may also show the following diagram.

Click Here To Show Diagram Code

[go]$$Bc Ideal extension
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . 4 . 2 . 1 . . 3 . . O . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . X . . . . . . . . . . . . . O . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |[/go]

If invades, it does not have room to make an ideal base, while does. Not that good players make that invasion, but this is why they don't.

Now, the bots do not play this way because they prefer other enclosures, but when they analyze pro games that do have these enclosures facing each other, they don't like . (I haven't seen everything, so sometimes they might, I dunno.)

First, they often play elsewhere instead of making this supposedly ideal move. IOW, maybe it's not so ideal. Second, this is the sequence they usually recommend, which we may consider a side joseki.

Click Here To Show Diagram Code

[go]$$Bc 3d line attachment first, high extension
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . 4 . . . |
$$ | . . . X . . . . . . . . . 3 1 O . . . |
$$ | . . . , . . . . . 5 . . . . 2 , . . . |
$$ | . . X . . . . . . . . . . . . . O . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |[/go]

Once you see this sequence, it makes sense. and support . - was played in the 19th century, but not in this situation. Furthermore, and hinder or squelch even the now tepid looking White development on the top side in an earlier diagram. It's not just that Black has a better development than White, what development by White? Why the 4th line extension instead of the 3d line extension? For one thing, the 3d line extension is vulnerable to a shoulder blow.

One more thing.

Click Here To Show Diagram Code

[go]$$Bc New White reply
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . . . . 5 1 4 a . . . O . . . |
$$ | . . . , . . . . 3 2 6 . . . . , . . . |
$$ | . . X . . . . . . . . . . . . . O . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |[/go]

The bots have shown us , something that humans overlooked in this situation. - is a natural sequence. Isn't this better for White than at a? and are oriented towards the side, and are oriented towards the center.



Not a bad idea.



That is something that I have often observed in SDKs. When they were DDKs they explored a lot of moves because they didn't know anything. But as SDKs they have learned something, and now they don't explore enough. They stick to what they know. The bots have shown us that even pros suffer from the same thing, if not to the same degree.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

I started to go through OTME (Opening Theory Made Easy) today. Lo and behold, the first principle Otake (9p) discussed was essentially the same idea that Bill suggests is wrong.



Now, my first instinct is that this situation is bad for white. Black is ahead and gaining more with each jump. But how much more? One line of territory, so 5 points at this stage + sente. One principle from this example, that neither player should put themselves in a similar situation, is sound. I.e. strive for an even exchange at least. What does Katago say?

Well, Katago, looking at the full board situation from the book, prefers to deal with the other corners first. One possible lesson is that corners are more urgent than sides, even when there is a relatively "hot" side. I decided to let the game continue from there, to see when Katago would cast its silver eye toward the right-hand side.



This position is still highly symmetrical. Katago slightly prefers the extension over the attachment in this situation. But, letting it ponder even further, we get:



Okay, this move also makes sense in this situation. But, all three moves (extension on the bottom, extension on the right, and attachment) are within a percentage point of each other. Judging between them is impossible for my skill level, but I suggest that any move that would only be a mistake 1% of the time with nearly perfect play cannot be a serious blunder. Indeed, a huge philosophical issue now comes up. Are moves that are good in an objective sense--for strong amateurs, pros, and machine gods--necessarily good strategy for weaker players? The attachment leads to a more complicated, less predictable game. It probably matters more now in which sort of game the player performs better.

Other thoughts: Perhaps Katago is saying something closer to: "All this moves have very similar values. Play to your strengths." In pro play, it might be a better strategy to deeply study some unusual yet sub-optimal opening and hope that your foreknowledge of it gives you the advantage. I recall Michael Redmond doing something similar to that in a game he reviewed on the AGA channel. His opponent purposely made a sub-optimal play designed to take the game away from variations that Redmond was familiar with.

A practical take away: It's probably okay to play in the corners before the sides, assuming that the corners are still unsettled and hot.

_________________
Now men awoke and listened to Felagund as he harped and sang, and each thought that he was in some fair dream, until he saw that his fellows were awake also beside him; but they did not speak or stir while Felagund still played

Because the white shimari is a high rather than low one, black's outside attachment is a relatively less good move, and the approach on the lower side undercutting the high stone is now a much better move.

LeelaZero taught us that the 2nd or even 3rd move in a corner is usually more valuable than a move on the side, see the absence of side moves in my opening gospel according to LZ. I'm not as familiar with KataGo but it seems to be aligned (as was Elf). One way we can understand this is that the 2nd move in a corner (shimari, 3-3 invasion, approach) or even the 3rd (atari and descend on the 3-3 invasions and jump josekis in your picture) changes the strength and weakness of groups, whereas the kind of side extensions favoured in the old theory expounded by Otake are purely territory potential expanding/denying moves, there's not a (potentially) weak group in sight.


You've definitely miscounted one thing here, and possibly 2, and neglected another. Black is building an area 5 lines wide in front of his shimari, and white has 3. 5 - 3 = 2, not 1. As for how many points each line is worth, I wouldn't count a box territory with a jump to 5th line as sure 4 points per line: there's classic reduction moves like the 5th line reduction which cut it down lower. But even more importantly black's territory is both wider and less secure. So if we do count it as 4 points per line, black has a 5 * 4 = 20 box vs white's 3 * 4 = 12. But you need to adjust these by the probability of them becoming territory, which I would hand-wavy estimate as 60% for black's and 90% for white, So a fairer comparison would be 20 * 60% = 12, vs 12 * 90% = 10.8.

I didn't mean to suggest that, only that the top bots prefer the corners over the sides more than humans have done. (OC, that doesn't mean that the bots are right. Top bots 10 years from now may have different ideas. )



I take it you are referring to the right side after both extensions and jumps. I expect that you are right. OC, White does not have to play that extension.




As I did explicitly suggest.



That's what I think, too. With enclosure facing enclosure, we have a rather hot side. I did label Black's extension as ideal, and for extensions, it pretty much is. I checked the book position with the Elf commentaries of GoGoD games. Kono Rin (B) vs. Mitsunaga Junzo, 2008-05-22s. has exactly that position, and Kono did play the book extension. Like KataGo, Elf prefers to invade the top left corner, a heresy when the book was written. At the same time, its preference is less than 2%, which is not enough for us to think that the extension is a mistake.

I decided to let the game continue from there, to see when Katago would cast its silver eye toward the right-hand side.


Bravo! You let KataGo run until its top choice had many rollouts. For analysis, I like to have at least 10k rollouts for each option under consideration.



Yes, this approach to the bottom right high enclosure goes back at least to the 19th century, as well. It's a good move to learn.



Well, yes. But this is an ideal, or nearly ideal extension, which is why it the traditional opening theory thinks that it is so good. But the bots regard it as pretty much run of the mill for this stage of the game.



Well, what is your objective? Is it to win games or to advance? (Obviously, the two are not mutually exclusive. If you win a lot of games, you will advance, and if you advance you will win a lot of games. ) You have to stretch your comfort level. Invest in loss, as the Tao Teh Ching suggests.

Anyway, it sounds like your idea of reading OTME while consulting KataGo is paying off.

Good luck!

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Another example from the Kono - Mitsunaga game

Click Here To Show Diagram Code

[go]$$Bcm15 Double wing and last big point of the opening
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . O . . . . . . . . X . . . . |
$$ | . . . O . . . . . 1 . . . . . , X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . 2 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , X . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . O . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . X . X . . . , . . . . O , O . . |
$$ | . . . . . . . . . X . . X . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

makes another ideal shape, the double wing from an enclosure, and in addition gets the last big stakeout play of the opening, something that traditional theory considers a plus. This combination is, if anything, more highly regarded by theory than the extension between facing enclosures. Mitsunaga does not play another stakeout play, but invades the right side. At this point you might consider that the game is in the early middle game or the late opening stage. There is no question that this extension is highly regarded by traditional opening theory.

However, Elf, with 25.5k rollouts for , regards it as a mistake, losing 8½% by comparison with its top choice (next diagram). OC, that choice is a corner play. Now Elf is highly opinionated, and the version used by the commentary is a year and a half old and weaker than KataGo and Lizzie now. Still, I would be surprised if Katago with a comparable number of rollouts made the double wing its top choice. I don't know about KataGo's winrate estimates, but maybe it will rate as around 4% worse than its top choice. Just a guess.

Click Here To Show Diagram Code

[go]$$Bcm15 Elf's mainline variation
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . O . . . . . . . . X . . . . |
$$ | . . . O . 4 3 . . , . . . . . , X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . 5 . . . . . , . . . . . , X . . |
$$ | . . O . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . O . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . O 2 . . . . . . . . . . . . . . . |
$$ | . . 1 . . . . . . . . . . . . . . . . |
$$ | . . . X . X . . . , . . . . O , O . . |
$$ | . . . . . . . . . X . . X . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

Elf likes the kick for (28k rollouts). Kono might have chosen that, as well. Then Elf likes the shoulder blow for (19.5k rollouts). (Note: The number of rollouts for plays in the variation are part of the rollouts for previous plays, and are therefore fewer in number. The moves are not independently generated.) Then another shoulder blow, , gets 8.2k rollouts. The thing is, Elf doesn't think much of the double wing or of getting the last big stakeout play. It prefers a couple of corner sente, followed by reducing the left side.

Edit: Kono's double wing play surrenders the initiative to Mitsunaga, who makes an invasion. Elf's plays for Black are more dynamic. All three put the question to White. This kind of attitude wins games against humans, as well.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Yeah, I probably misinterpreted your post. I still don't understand what traditional opening theory has gotten wrong since I don't understand what it is. If it's just a set of good moves and bad moves, then it's not so much a theory as a disconnected opinions. From OTME, it seems like the idea in this classic example is that regions between white's zone of influence and black's zone are especially valuable, since play in those regions both expands your own territory and restricts your opponent's.






Thanks. It was almost by accident though, since I came back to the process and noticed that the preferred option had changed.



Since I measure advancement by winning games, I would never be able to tell if I some how won more games while not actually advancing. Therefore, I do not distinguish between the two objectives.



We'll see. It does make reading the book more time consuming.



That's a tasty bit of theory.




Like I mentioned before, I don't have much experience with positional judgement. Just a binary feeling: good or bad. I was trying to compare value of black's whole moyo with white's whole moyo, but I did simplify by not considering how secure the moyo could be, mainly because I do not even have a "hand-wavy estimate" of that. I guess this isn't the best way to find the value of a specific move, like the extension or either of the jumps.

Maybe just posting slightly to very wrong ideas here is even more productive than reading old books, since the explanations I have been getting here are extremely thorough. Thanks so much for these little articles.

_________________
Now men awoke and listened to Felagund as he harped and sang, and each thought that he was in some fair dream, until he saw that his fellows were awake also beside him; but they did not speak or stir while Felagund still played

Just want to say that even though I can't contribute, I'm enjoying the combination of AI review of opening theory and discussion of its implications. Thanks for all the quality posts.