It is currently Thu Mar 28, 2024 11:40 am

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 16 posts ] 
Author Message
Offline
 Post subject: A close game against a 4k, review please
Post #1 Posted: Sat Nov 18, 2017 1:00 am 
Lives with ko

Posts: 155
Liked others: 160
Was liked: 4
Rank: 4k OGS 1d Fox
Online playing schedule: OGS, Fox Server
Hi all,

I played an even game against a 4k kyu, while I am now a 6 kyu. I am white: putting aside my final blunder, can you please give me a review? thanks in advance


_________________
Don't play 1-2-3
Just play 3

(Go proverb)

Top
 Profile  
 
Offline
 Post subject: Re: A close game against a 4k, review please
Post #2 Posted: Sat Nov 18, 2017 10:44 pm 
Dies with sente

Posts: 75
Liked others: 10
Was liked: 8
Rank: SDK
I'm around your rank, so I can't really offer too much, but what I did notice is that by W130, you seem to be well ahead in territory before komi. The only significant territory B had was around the top and the right area and he needed to convert most of the center moyo into territory to catch up.

I'm not sure if this is why you played W132 at N12, but I thought it was a bit slack. I think that reducing B's potential territory is much bigger. Initially I was thinking of a 1 space jump at J9, but I asked Leela, and it suggests F7. Whatever the case, we can see that White is poking out at the bottom left, the right, and has a huge entrance at the top center, so it should be easy to reduce Black. However, B ended up converting just about everything to territory.

EDIT: Changed F17 to F7


Last edited by WindCaliber on Mon Nov 20, 2017 11:28 am, edited 1 time in total.

This post by WindCaliber was liked by: Fllecha
Top
 Profile  
 
Offline
 Post subject: Re: A close game against a 4k, review please
Post #3 Posted: Sun Nov 19, 2017 1:48 am 
Lives with ko

Posts: 155
Liked others: 160
Was liked: 4
Rank: 4k OGS 1d Fox
Online playing schedule: OGS, Fox Server
F17 is occupied :-? you meant f7?

Yes that move was slack: the intent was to make a threat to the adjacent stone but that was smaller..

_________________
Don't play 1-2-3
Just play 3

(Go proverb)

Top
 Profile  
 
Offline
 Post subject:
Post #4 Posted: Sun Nov 19, 2017 2:28 am 
Honinbo
User avatar

Posts: 8859
Location: Santa Barbara, CA
Liked others: 349
Was liked: 2076
GD Posts: 312
Hi Fllecha,

:w6: (*) Between o3 and o17, I prefer o17;
but 3-3 (either one) also possible.

:w12: (*) I prefer E6.

:b13: Strange. E6 atari natural.

:w14: E6. ( :b13: missed this shared vital point. )

:w16: Very strange.

:b17: C14. I don't understand :w16: or :b17: . :-?

:w22: C14.

:w34: R13.

:b35: R15. Huge loss for W.

:w36: R13.

:b37: R15. Huge loss for W.

Understanding ( :w34: at R13 ) and ( :b35: push through at R15 ) is worth this game, IMO. Too fundamental. Thanks.

_________
(*) Curious about Leela/AGZ's evaluations.


This post by EdLee was liked by: Fllecha
Top
 Profile  
 
Offline
 Post subject: Re: A close game against a 4k, review please
Post #5 Posted: Sun Nov 19, 2017 3:07 am 
Judan

Posts: 6725
Location: Cambridge, UK
Liked others: 436
Was liked: 3719
Rank: UK 4 dan
KGS: Uberdude 4d
OGS: Uberdude 7d
The nice thing about F7 is it not only reduces the centre, it also applies some pressure to black's group on the left side which is not yet alive and has some cuttings points for future fun. By hassling this group you will be able to naturally trash black's centre potential, probably in sente.


This post by Uberdude was liked by: WindCaliber
Top
 Profile  
 
Offline
 Post subject: Re: A close game against a 4k, review please
Post #6 Posted: Sun Nov 19, 2017 5:30 am 
Oza
User avatar

Posts: 2408
Location: Ghent, Belgium
Liked others: 359
Was liked: 1019
Rank: KGS 2d OGS 1d Fox 4d
KGS: Artevelde
OGS: Knotwilg
Online playing schedule: UTC 18:00 - 22:00
I want to highlight one good and one bad thing about your game. Let's start with the bad:

You seem to be envious of what your opponent has and invade everywhere, which makes you end up with weak groups all over the place.

Example moves: 16, 34, 54

by 81 Black has effectively separated your groups into 7 apiece. This is strategically unsound.

However, the combination 98-100 is wonderful and shows you have some fine tactics up your sleeve and you are either audacious with good intuition or can read well ahead in fights.

At 103 I make a positional judgment how to carry this game to a comfortable victory. The timing is good to remember: after winning a local battle, look at the whole board.

(minor uncommented suggestions along the way)



This post by Knotwilg was liked by 2 people: emeraldemon, Fllecha
Top
 Profile  
 
Offline
 Post subject: Re:
Post #7 Posted: Mon Nov 20, 2017 10:11 pm 
Dies with sente

Posts: 75
Liked others: 10
Was liked: 8
Rank: SDK
EdLee wrote:
Hi Fllecha,

:w6: (*) Between o3 and o17, I prefer o17;
but 3-3 (either one) also possible.

:w12: (*) I prefer E6.

_________
(*) Curious about Leela/AGZ's evaluations.


I input the positions to Leela, and it pretty clearly favors O17 over O3. However, after a while it started favoring the ogeima enclosure at F17.

For :w12:, Leela also clearly prefers E6 over C5.

Top
 Profile  
 
Offline
 Post subject:
Post #8 Posted: Tue Nov 21, 2017 2:28 am 
Honinbo
User avatar

Posts: 8859
Location: Santa Barbara, CA
Liked others: 349
Was liked: 2076
GD Posts: 312
Hi WindCaliber,

Thank you. How did Leela's winrate change for :b5: ?
For :w6:, how did Leela feel about either 3-3 ?

Thanks.

Top
 Profile  
 
Offline
 Post subject: Re:
Post #9 Posted: Tue Nov 21, 2017 12:30 pm 
Dies with sente

Posts: 75
Liked others: 10
Was liked: 8
Rank: SDK
Leela prefers :b5: at E16.
Image

Leela does not like sanrensei, it seems. The winrate drops a few percentage points.
I realized I still had Leela 0.10 during my previous post, so I retested with Leela 0.11. It still prefers :w6: at O17 over O3, but it has some other opinions on the top left enclosures.
Image

Leela does not seem to prefer 3-3 invasion(I've never seen it play it, either.). I forced it to play 3-3, and it seems to think that it is slightly good for B.
Image
Image

Top
 Profile  
 
Offline
 Post subject: Re: A close game against a 4k, review please
Post #10 Posted: Tue Nov 21, 2017 12:40 pm 
Lives with ko

Posts: 155
Liked others: 160
Was liked: 4
Rank: 4k OGS 1d Fox
Online playing schedule: OGS, Fox Server
Hi all,

of course many thanks for helping me. I now also discovered Leela that is an incredible engine.. wow beautiful!!

I now have a better understanding of this game, and the blunder pointed by Ed lee: I accepted too play even game against a stronger player and I had better position but eventually I lost and that was hard to accept.

I am playing another two games and in a couple of days I'll post it: it's about moyo and it is a very important topic for me. I think I'm winning one and losing the other but we wait and see... thank you again :bow: :bow:

_________________
Don't play 1-2-3
Just play 3

(Go proverb)

Top
 Profile  
 
Offline
 Post subject:
Post #11 Posted: Tue Nov 21, 2017 12:48 pm 
Honinbo
User avatar

Posts: 8859
Location: Santa Barbara, CA
Liked others: 349
Was liked: 2076
GD Posts: 312
Hi WindCaliber, Thanks very much.
Quote:
Leela does not like sanrensei, it seems. The winrate drops a few percentage points.
Maybe Leela and AG agree with each other somewhat. (Maybe AG's % drops even more.)
Quote:
Leela does not seem to prefer 3-3 invasion. I forced it to play 3-3, and it seems to think that it is slightly good for B.
Hmm, maybe Leela is not quite AG's level. :) Will be curious to see AG's evaluations some day, if possible.

Thanks.

Top
 Profile  
 
Offline
 Post subject: Re:
Post #12 Posted: Tue Nov 21, 2017 6:13 pm 
Lives in gote
User avatar

Posts: 392
Liked others: 23
Was liked: 43
Rank: NR
EdLee wrote:
Maybe Leela and AG agree with each other somewhat.
Leela 11 (running on a dual-core CPU for a few minutes) agrees with Alfie (using 4 TPUs) at move 114 in game 10 of the AlfieM v AlfieM series - but i like Michael Redmond's quieter and more solid move better (a sound honte move which Leela didn't even consider).
https://papers.ssrn.com/sol3/papers.cfm ... id=3071677

In Filecha's game, i thought black 33 to be a grave error, as it induces (should have induced) white 34 at R13 as Ed points out, which as well as making white safe, ruins black's potential on the right and removes the aji of a possible black R12 later, which would create a defect in white at S15, so white would have to defend, allowing black R12 to make points in sente.

As for white 34, i feel that would be horrendous even if white already had R13 in place. Horrendous because (a) it's too close to black's wall; (b) it induces black to make herself even stronger at the top which is likely to end with black having sente (c) a much safer way to erase black's top would be a high approach, maybe at K15, which could run away lightly and start to get a white presence in the centre, or perhaps an alfie-like shoulder hit instead - M16 just creates a target for black; and (d), most of all, it's Wrong Direction, because white's bottom is wide open. [admin] Violation of TOS removed. -JB [/admin]

_________________
i shrink, therefore i swarm

Top
 Profile  
 
Offline
 Post subject: Re: Re:
Post #13 Posted: Tue Nov 28, 2017 2:06 am 
Lives with ko

Posts: 155
Liked others: 160
Was liked: 4
Rank: 4k OGS 1d Fox
Online playing schedule: OGS, Fox Server
djhbrown wrote:
most of all, it's Wrong Direction, because white's bottom is wide open...


:D :D hhaha thanks ;)

Here what i was thinking at that point:

"I am unconfortable because in spite of my three corners I have to face that upper moyo. Usually I lose when I an not able to destroy it." So three moves came to my mind:

a) K15-like moves (which is probably better standard reducing move tham my first candidate K13) were in my feelings too slow, and too much passive: black can simply ignore and go on with tengen moves and I have to survive and avoid a central wall

b) Moves like K4. Obv that was first idea. Gote but I'm not wide open... but easy to invade and make some jeopardy, hard to handle.

c) I considered finally the move I played, M16, which was like a compromise: probably eventually is gote, but better reducing move, more annoying and pseudo-sente, at least for the moment, and I create a chance to make my opponent panic-err and maybe take the initiative. And maybe later try to close the bottom.

Of course Ed Lee's move never passed close to my mind.

This is NOT meant to be a correct thought, of course, so any correction is appreciated ;)

_________________
Don't play 1-2-3
Just play 3

(Go proverb)

Top
 Profile  
 
Offline
 Post subject: Re: Re:
Post #14 Posted: Wed Nov 29, 2017 4:27 am 
Lives in gote
User avatar

Posts: 392
Liked others: 23
Was liked: 43
Rank: NR
Fllecha wrote:
I have to face that upper moyo. Usually I lose when I an not able to destroy it.
Perhaps you can change the way you think about moyos - you don't have to destroy a moyo to win.

Maybe Swim's view can help you assess the situation?
Attachment:
flecha.png
flecha.png [ 61.59 KiB | Viewed 9796 times ]

There's a big hole in white's moyo at the bottom, and the centre is shadowed on 3 sides by black. So it would be easy for black to invade white and escape.

There's a small hole in black's moyo, but it's not big enough for white to live in (the smallest living group needs an empty area of about 7x5 or 7x3 on the side). And because black has more centre power than white, it will be difficult for white to escape if she invades black too deeply. And white's nearest friends are very far away.

So, invading black deeper than K10 is going to be tough (even K15 is too deep).

But white can't even play K10 just yet, because if she did, black could attack it from underneath at K8, making a follow-up black invasion of the bottom even more severe.

White's bottom right stones aren't alive yet; the usual joseki has white playing at L3 as well to make a secure position.

White can live in the top right without R13, but right now white's best target is black's right side... at the moment it's very loose, but if black got to play R12 first, that would change the picture a lot...
Attachment:
f2.png
f2.png [ 62.44 KiB | Viewed 9796 times ]

_________________
i shrink, therefore i swarm


This post by djhbrown was liked by: Fllecha
Top
 Profile  
 
Offline
 Post subject: Re: A close game against a 4k, review please
Post #15 Posted: Wed Nov 29, 2017 10:15 am 
Lives with ko

Posts: 155
Liked others: 160
Was liked: 4
Rank: 4k OGS 1d Fox
Online playing schedule: OGS, Fox Server
djhbrown, thanks for the diagrams: put that way is easy to see white weaknesses, and thank for the thinking process, iit helped a lot. One final question: what is "Swim"? is a go program/ go diagram generator? Couldn't find anything on google :-?

_________________
Don't play 1-2-3
Just play 3

(Go proverb)

Top
 Profile  
 
Offline
 Post subject: Re: A close game against a 4k, review please
Post #16 Posted: Wed Nov 29, 2017 4:49 pm 
Lives in gote
User avatar

Posts: 392
Liked others: 23
Was liked: 43
Rank: NR
Fllecha wrote:
what is "Swim"?
Swim = See what i mean - it's my retirement hobby design of a Go playing algorithm that "thinks" and "talks" about Go in a commonsense way (ie in what i think is a commonsense way, based on what i've read about such things in the AI and cognitive psychology literature - as i write this, i recall that one of the things that most caught my imagination in 1971 was a book called "Remembering" by F. Bartlett).

Ironically, i can't remember now what Bartlett says! Luckily, Google has a better memory than me:
https://www.google.com.au/url?sa=t&rct= ... NeBCE3W3TS

In 2015, (a few months before Alphago burst on the scene), having unravelled the jigsaw of God (she started off 37000 years ago as a woman, but was changed into a man by the agricultural revolution about 8000 years before The Bible was written):
https://www.youtube.com/watch?v=5lsQUq9 ... Q2h2UALN0h
and unable to think of anything better to do, i started exploring what a professional player sees when they look at a Go position. One thing led to another, and i ended up making about 35 movies about it, in the playlist:
https://www.youtube.com/watch?v=Om_CVAe ... a3Hl1X_v-S

My latest effort is on how Swim can learn new tricks:
https://papers.ssrn.com/sol3/papers.cfm ... id=3071677

The pictures you saw are produced by "Gomap", a prototype implementation by pnprog of Swim's basic perceptions of connectedness and influence:
https://github.com/pnprog/gomap

Hopefully, one day, someone with more energy and programming skill than me will have a Go at implementing a bit more of Swim, so it can be proven beyond any doubt that it could beat Alphago zero...

Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 16 posts ] 

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group