Posts: 10660Liked others: 3586 Was liked: 3337
Discussion of Elf's commentary
In the Elf commentaries Elf, assuming area scoring and 7½ pts. komi, estimated the winrate for one player as 90½% (58.5k rollouts) after
. Which player was it?
Black . And Black won by resignation.
Click Here To Show Diagram Code [go]$$Wcm32 Black is aheadThis was a game between Honinbo Shusai Meijin (W) and Segoe Kensaku, 5 dan, GoGoD 1920-01-17a. OC, it was a no komi game.
What now?
Click Here To Show Diagram Code [go]$$Wcm32 Game continuation -
are all Elf's top choices. Shusai is seeking complications. Elf's top choice for
is
a , but it gives Shusai's hane a slightly better winrate for White. Both moves lead to complicated fights. It seems like the main idea of
a is to bolster the White wall by sacrificing the other White stones in the top right. The next diagram shows Elf's mainline after Shusai's hane.
Click Here To Show Diagram Code [go]$$Bcm37 Elf's mainline variation for makes a solid connection and then
saves his group on the right side. Then
attacks the White wall.
plays kikashi against the corner, and then
attaches for sabaki.
pulls back, preventing a hane or crosscut there.
runs out towards the center.
Earlier in the game
Click Here To Show Diagram Code [go]$$Wcm6 Joseki loses 12½% to parThe extension,
, is joseki and it checks Black's development on the left side, in particular a pincer. I don't think that this play would have come in for criticism before the AI era.
Click Here To Show Diagram Code [go]$$Wcm6 Elf's mainiline variation for White plays a bot favorite, the press. After the usual continuation,
occupies the last open corner, a kind of tedomari.
Click Here To Show Diagram Code [go]$$Bcm7 Game continuation -
may be best in a no komi game, but Elf, assuming a 7½ komi, docks
11½% and
10%. For both plays Elf prefers the enclosure at
a , which prevents the 3-3 invasion.
Elf's mainline variation for
Click Here To Show Diagram Code [go]$$Wcm8 Two enclosures is unappetizing in a no komi game, as it allows Black to make two enclosures. Black has an easy opening, now.
plays a taisha against
. Before the AI era White could hardly have predicted the slide at
, which seems passive.
approaches the high enclosure in the bottom left and
pincers.
is interesting, a jump instead of making a base at
a . It is not as though Black has a strong attack against
and
.
Elf's mainline variation for
Click Here To Show Diagram Code [go]$$Wcm12 Black attackAfter
,
pincers White's two stones in the top right corner. It goes against the grain, at least psychologically, for White to be attacked by Black instead of attacking Black in a no komi game. Again, Black has an easy opening. But these diagrams illustrate the value Elf places on the bottom right enclosure.
Click Here To Show Diagram Code [go]$$Bcm13 Game continuation -
are all Elf's top choices. I find
of some interest. OC, it is joseki, but one strategy for White, it seems to me, going back at least as far as Dosaku, is to sketch out a very loose sphere of influence, which we cannot even call a moyo. If Black comes in, he can be attacked. In this game White eventually made a moyo, which Black invaded — twice.
Elf thinks that
loses 9½% to the more solid play at
a .
is an anti-shape play, preventing a Black block there, but Elf thinks that it loses 14% to par.
is a creative idea, but Elf docks it 15% versus the solid connection at
a and the 3-3- invasion at
b . Elf regards
as White's largest mistake so far, losing 16½% to par. White's play here seems altogether too passive. In the top right, Elf thinks that
and
are minor errors, that each one would have been better at
c . With
perhaps White was pursuing the strategy I mentioned before.
Elf's mainline variation for
Click Here To Show Diagram Code [go]$$Wcm18 Mainline variation for at
Shusai would have seen this sequence in a instant. One problem is that it weakens the
stones.
Click Here To Show Diagram Code [go]$$Bcm27 Mainline variation for White makes nice territory in exchange for Black's attack. Good for White, says Elf. (Black's top left group has problems, too.) This diagram shows why Elf did not like Black's jump to
.
Elf's mainline variation for
Click Here To Show Diagram Code [go]$$Bcm19 Mainline for =
,
=
is usual, forcing
. Then Black jumps into the 3-3 point in the bottom right corner. Note:
at
transposes to this position, as well.
Elf's mainline variation for
Click Here To Show Diagram Code [go]$$Wcm20 Mainline variation for and
push through and cut.
does not go quietly, but makes a fight of it.
threatens the Black corner group. Black does not want White to play at
27 .
Click Here To Show Diagram Code [go]$$Wcm30 Mainline variation for connects his stones, and then the sharp fighting continues.
Click Here To Show Diagram Code [go]$$Wcm40 Mainline variation for at
is a nice sacrifice tesuji.
After
Black has made thickness, but White has taken a good bit of profit.
Elf's variations rarely last so long, especially since
got only 29.6k rollouts. But it is pretty much a one lane road, with few rollouts lost per move to alternatives.
----
White lost ground in the opening, which is normal in no komi games, but he blew more than 30% in two moves in the top left, by not being aggressive enough.
At some point, doesn't thinking have to go on?
Login
Register
FAQ
Search
Bill Spight
. Which player was it?
- 
are all Elf's top choices. Shusai is seeking complications. Elf's top choice for 
is a, but it gives Shusai's hane a slightly better winrate for White. Both moves lead to complicated fights. It seems like the main idea of a is to bolster the White wall by sacrificing the other White stones in the top right. The next diagram shows Elf's mainline after Shusai's hane.
saves his group on the right side. Then 
attacks the White wall. 
plays kikashi against the corner, and then 
attaches for sabaki. 
pulls back, preventing a hane or crosscut there. 
runs out towards the center.
, is joseki and it checks Black's development on the left side, in particular a pincer. I don't think that this play would have come in for criticism before the AI era. 
occupies the last open corner, a kind of tedomari. 
- 
may be best in a no komi game, but Elf, assuming a 7½ komi, docks 
. Before the AI era White could hardly have predicted the slide at 
, which seems passive. 
pincers. 
is interesting, a jump instead of making a base at a. It is not as though Black has a strong attack against 
.
are all Elf's top choices. I find 
loses 9½% to the more solid play at a. 
is an anti-shape play, preventing a Black block there, but Elf thinks that it loses 14% to par. 
is a creative idea, but Elf docks it 15% versus the solid connection at a and the 3-3- invasion at b. Elf regards 
as White's largest mistake so far, losing 16½% to par. White's play here seems altogether too passive. In the top right, Elf thinks that 
and 
are minor errors, that each one would have been better at c. With 
at 
stones.
, 
does not go quietly, but makes a fight of it. 
threatens the Black corner group. Black does not want White to play at 27.
at 
After