How is your positional judgement? Game 20

Bill Spight · #1

Assuming area scoring with 7.5 pt. komi, after :b27:

Elf estimates a winrate of 90% (49.1k playouts) for one player. Which player is it?

Click Here To Show Diagram Code: [go]$$Bc White to play $$ --------------------------------------- $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . X . . . . . . . . . . O X . . | $$ | . . O , . X . O . , . O . . . O X . . | $$ | . . . . . . . . . . . . . . . O X . . | $$ | . . . O . . . . . . . . . . . . X . . | $$ | . . . . . . . . . . . . . . O . . . . | $$ | . . . . . . . . . . . . . . X X . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . O , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . O O . . | $$ | . . . . . . . . . . . . . . X . X X . | $$ | . . O , . . . . . , . . . . . , . . . | $$ | . . . . O . . . . X . . . . . X . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ ---------------------------------------[/go]

Enjoy!

Knotwilg · #2

Tryss · #3

jlt · #4

Kirby · #5

bernds · #6

Bki · #7

TelegraphGo · #8

xela · #9

Interesting, this one I feel strongly that I should stay with my "at a glance" assessment and reading won't help (in contrast to that pesky game 19). At least, not for my reading ability :-)

OK, now that I've tried thinking for myself, let's see what some bots think.

Knotwilg · **#10**

Some variations of mine, after seeing my judgment was wrong, and one variation by LZ:

Uberdude · **#11**

I think whole game tewari with my Elf hat on is a better judge than static positional judgement so here is a plausible sequence of play with my Elvish commentary.

Top right has the same weird "was it a shoulder hit to 3-3 and they blocked but then didn't hane" weird shape from game 12, in which Bill revealed it was the result of an initial 6-4 stone so let's say it was that. So Elf thinks 1 is quite bad (-12% before), 2-4 ok, 5 and 6 a little slow but equally so. 7 is good, bots like low approach to 3-4 (but probably closing top right more, maybe that corner actually happened before top left). 8 locally a bit bad (prefer to defend on left to prevent press) and globally more so as entering top left bigger, but then 9 is worse (press or tenuki better), 10 is ok. So black lost in this corner so far

Click Here To Show Diagram Code: [go]$$Bc $$ +---------------------------------------+ $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . 7 . . . . . . . . . . . . . . | $$ | . . 2 , . 9 . 8 . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . 0 . . . . . . . . . . . . 1 . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . 5 . . | $$ | . . 4 , . . . . . , . . . . . , . . . | $$ | . . . . 6 . . . . . . . . . . 3 . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ +---------------------------------------+[/go]

11 is definitely smaller than top right shimari, and locally probably slow this early in opening (but maybe it was later). 12 is too slow, and side move which bots don't like, 13 likewise, but probably a bit worse as white's side move does help defend against a weakness of group at top left (now that black slid to take corner) whereas black's is pure points. But now at least I have a white sente to go into the top right. In the last game both players made fair sized mistakes in this sequence, but the net effect can be simplified as the sticks of 3 stones are basically even but then black's r14 should be a hane not extend so let's say -10%+.

Click Here To Show Diagram Code: [go]$$Bcm11 $$ +---------------------------------------+ $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . 1 . . . . . . . . . . . . . . . . | $$ | . . . . X . . . . . . . . . . 4 5 . . | $$ | . . O , . X . O . , . . . . . 6 7 . . | $$ | . . . . . . . . . . . . . . . 8 9 . . | $$ | . . . O . . . . . . . . . . . . X . . | $$ | . . . . . . . . . . . . . . 0 . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . 2 , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . X . . | $$ | . . O , . . . . . , . . . . . , . . . | $$ | . . . . O . . . . 3 . . . . . X . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ +---------------------------------------+[/go]

20 for 21 is then a good exchange for white, event though black is building 4th line territory, 21 is just a points move as the 4 stones are already alive (white's follow up at r14 is gote, has cuts and black corner still happy), whereas 20 is important for strength of white's stones and avoiding too much shortage of libs pain. Would a pro really play 21 though rather than invade the top and attack the wall? 23 then slow? 24 a shoulder hit reduction a bot classic, 25 is a weird move we don't see bots play so presumably bad, 26 for 27 a nice little exchange because 25 didn't add libs to r5, and now white has sente and knight move at o7 to run away looks reasonable. So white leading.

Click Here To Show Diagram Code: [go]$$Wcm20 $$ +---------------------------------------+ $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . X . . . . . . . . . . . . . . . . | $$ | . . . . X . . . . . . . . . . O X . . | $$ | . . O , . X . O . , . 3 . . . O X . . | $$ | . . . . . . . . . . . . . . . O X . . | $$ | . . . O . . . . . . . . . . . . X . . | $$ | . . . . . . . . . . . . . . 1 . . . . | $$ | . . . . . . . . . . . . . . 4 2 . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . O , . . . . . , . . . . . , . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . 5 7 . . | $$ | . . . . . . . . . . . . . . 6 . X 8 . | $$ | . . O , . . . . . , . . . . . , . . . | $$ | . . . . O . . . . X . . . . . X . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ | . . . . . . . . . . . . . . . . . . . | $$ +---------------------------------------+[/go]

Kirby · **#12**

It’d be an interesting stat to see how the actual game results correlated with Elf’s assessment, aggregated across all of these games.

These are all pro games, so I wonder how often the end result matched up such that Elf’s favored player won.

Bill Spight · **#13**

In this case the player favored by Elf won the game.

Kirby · **#14**

Bill Spight wrote:

In this case the player favored by Elf won the game.

Whoops - I guess I’m posting in the wrong thread (should’ve been the main one)..

Bill Spight · **#15**

Kirby wrote:

Bill Spight wrote:

In this case the player favored by Elf won the game.

Whoops - I guess I’m posting in the wrong thread (should’ve been the main one)..

That's OK. Message received.

I'll take care of it.

Knotwilg · **#16**

Uberdude wrote:

I think whole game tewari with my Elf hat on is a better judge than static positional judgement so here is a plausible sequence of play with my Elvish commentary.

That is how I have understood, from John and Bill, how professionals think. I find it a very interesting perspective that historical evaluation would give better insight than static evaluation and I'm impressed you can do it so well.

Uberdude · **#17**

Knotwilg wrote:

Uberdude wrote:

I think whole game tewari with my Elf hat on is a better judge than static positional judgement so here is a plausible sequence of play with my Elvish commentary.

That is how I have understood, from John and Bill, how professionals think. I find it a very interesting perspective that historical evaluation would give better insight than static evaluation and I'm impressed you can do it so well.

It works best with this kind of peaceful fuseki-style opening in which I can apply the bot-style opening principles I've absorbed over the last few years so have reasonable confidence in my ability to predict whether Elf thinks moves are good or bad. I'm much less able to do that on more dynamic fighting positions like game 19.

Bill Spight · **#18**

Knotwilg wrote:

Uberdude wrote:

I think whole game tewari with my Elf hat on is a better judge than static positional judgement so here is a plausible sequence of play with my Elvish commentary.

That is how I have understood, from John and Bill, how professionals think. I find it a very interesting perspective that historical evaluation would give better insight than static evaluation and I'm impressed you can do it so well.

I don't know about other professionals, but Takemiya has said that he relies upon shape, overconcentration, etc., because with his style of play he cannot rely upon counting territory. At least, early in the game.

Bill Spight · **#19**

Kirby wrote:

It’d be an interesting stat to see how the actual game results correlated with Elf’s assessment, aggregated across all of these games.

These are all pro games, so I wonder how often the end result matched up such that Elf’s favored player won.

I have shown the actual game winner in my discussion of all games through Game 17, which is as far as I have caught up with my discussions.