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.FinrodFelagund wrote: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.
)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.Now, my first instinct is that this situation is bad for white. Black is ahead and gaining more with each jump.
What does Katago say?
Well, Katago, looking at the full board situation from the book, prefers to deal with the other corners first.
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.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.
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.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:
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.Okay, this move also makes sense in this situation.
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.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.
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.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.
) 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!
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.
(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. 
As far as I kow the split has pretty much vanished in high level play because the other player is often the one choosing the direction of the checking extension, thus your opponent can use his initiative first to play elsewhere and return at a more appropriate time. I'd approach White's komoku.
Yep, the first time a bot recommended this extension, I was also mildly baffled. Feeling-wise I'm still not accustomed to it but I play it, too, all the time.
First feeling would be 3*3. Second feeling would also be the lower side (widest open area). You have a solid two-space-extension on the left and access to the corner. There is no weakness.
My first feeling is shoulder-hit on G4. But that might turn out bad because the left has some standard invasion points. Two-point-jump, second feeling. Third feeling G3 attachment, because jumping does only run and not separate - settling seems better.
In my opinion less to do with tactics but with pattern recognition.