| Life In 19x19 http://www.lifein19x19.com/ |
|
| Slonection http://www.lifein19x19.com/viewtopic.php?f=15&t=18047 |
Page 1 of 2 |
| Author: | Knotwilg [ Wed Feb 10, 2021 6:35 am ] |
| Post subject: | Slonection |
Coining a term in jest, if there's any pattern in the 40 games I played and reviewed over the past 40 days, then it's KataGo telling me I'm guilty of playing slow connections. It's so deeply ingrained in my system and probably in many others', that it becomes a heuristic of its own. Now I'll leave the enrichment of the English (Go) language to those qualified and "slonection" doesn't appeal to me any more than "anti-honte" does. Here's a diagram to start the thread: Following White's marked move I considered A to E (clockwise). What would you play? I didn't like any of those, so I looked for and found a more active sequence, which I was very satisfied with. However, KataGo didn't like it at all. I thought sealing in White was *awesome*. KataGo thought giving unconditional life to an opponent in gote, with cutting points in the seal, is a bad result. White takes sente in the lower right. |
|
| Author: | Knotwilg [ Wed Feb 10, 2021 6:44 am ] |
| Post subject: | Re: Slonection |
In the same game, I played not a slow connection but slow life: In my reading, I needed to add another move, because White can kill at A. If White answers, I live in sente, if not I can connect with points later at B. KataGo thinks I can live and also in some of the variations sacrifices the corner. This is another pattern in KataGo analyses: living in the corner in gote can be delayed in favor of bullying the surrounding group. At worst, the opponent spends many moves capturing. |
|
| Author: | Knotwilg [ Wed Feb 10, 2021 6:52 am ] |
| Post subject: | Re: Slonection |
A reduction of the lower side resulted in a chase to the center. White cuts at  . While  is a sharp move, with  I fall into the honte-trap again, playing a slow connection. I should have continued playing sharply at A.Another slow connection followed a few moves later: Instead: Black should cut in sente and then connect. Connecting first reduces the value of this sequence. |
|
| Author: | jlt [ Wed Feb 10, 2021 7:08 am ] |
| Post subject: | Re: Slonection |
In your first diagram, I like a. Either White takes the corner and lets Black escape on the side, then K16 is weakened, or White lets Black live in the corner and takes influence, but influence doesn't do much because Black has many stones on the fourth line already. I don't like b. This is an extension from an already living group. It has a followup, but if White simply answers, then this allows White to strength K16 and make a similar number of points at the same time. I don't like c for the same kind of reason. It does make the corner more difficult to invade, but I don't think it prevents invasion completely. "d" looks a good reduction move, not too close from other black stones. "e" somehow repairs the weakness at C7 but looks slow. So my order of preference is a, d, c, b, e. The probability that I am correct is 1/5! = 1/120. I you third diagram: in general I understand the concept of bullying the surrounding group before living, and I did notice that when reviewing my games with Katago, however in you diagram the surrounding group looks so strong that I don't see how bullying can be possible. |
|
| Author: | dfan [ Wed Feb 10, 2021 7:21 am ] |
| Post subject: | Re: Slonection |
Knotwilg wrote: Coining a term in jest, if there's any pattern in the 40 games I played and reviewed over the past 40 days, then it's KataGo telling me I'm guilty of playing slow connections. It's so deeply ingrained in my system and probably in many others', that it becomes a heuristic of its own. Now I'll leave the enrichment of the English (Go) language to those qualified and "slonection" doesn't appeal to me any more than "anti-honte" does. I feel like your examples all fit into the "slack moves" category of Go sins. As you allude to, it's sort of being "too honte". I'm very guilty of this (it's the main thing In-seong criticizes me for), and I think there are two principal sorts of reasons for it:
|
|
| Author: | Boidhre [ Wed Feb 10, 2021 8:19 am ] |
| Post subject: | Re: Slonection |
For me it's been "where is the big place" being repeated over and over in streams/videos I've watched and trying to tear my attention away from the local position when it's not longer urgent. Playing on the bottom first at c or d or playing in the top left corner at a (or similar moves) would be first instinct looking at the board and deciding I don't need to add another move in the top right to live and white approaching on top is easier to ignore than white approaching on the bottom. |
|
| Author: | pwaldron [ Wed Feb 10, 2021 12:55 pm ] |
| Post subject: | Re: Slonection |
Knotwilg wrote: Here's a diagram to start the thread: Following White's marked move I considered A to E (clockwise). What would you play? I'll take c > d > a > b > e. To the lower side looks important because there are good extensions for both sides, and both sides have good follow-ups afterwards. Black has the large shimari, which can be thin but can also look good when combined with a developing/reinforcing move. 'b' just seems to be a territory move, so I like 'a' better. I don't like 'e'. It's not obviously creating a weak group and it gives White opportunity to reinforce the corner to something large. Seems like a poor trade, even assuming there's nothing better available for White. Quote: I'm with you on this one. I think I'd play in the corner, and even with your hint I still don't see the theme of anything better. On a different topic, do you feel that Black's group on the left side is a little overconcentrated? There are three extra stones there compared to the earlier diagram, and to me it doesn't feel like Black's position improved as much as White's. |
|
| Author: | Bill Spight [ Wed Feb 10, 2021 2:17 pm ] |
| Post subject: | Re: Slonection |
In such positions, this late in the opening, I now ask myself, Where would Shusai play? Before the AI era, I would probably have chosen d, because c invites the jump to g, which looks too good for White. But I don't think that that would have bothered Shusai. So c is now my top choice. Second place is between d and f. a is too close to the corner. We don't want the  extension to look good. I think that f is the play to exploit the thinness of White's top side. So that's my second choice. If White takes sente and jumps to g, Black can still extend from bottom right corner, and if not, Black can reduce White's bottom side.
|
|
| Author: | John Fairbairn [ Wed Feb 10, 2021 3:03 pm ] |
| Post subject: | Re: Slonection |
Quote: In such positions, this late in the opening, I now ask myself, Where would Shusai play? Before the AI era, I would probably have chosen d, because c invites the jump to g, which looks too good for White. But I don't think that that would have bothered Shusai. So c is now my top choice. When commenting on other players' games, Shusai indeed VERY often tells them they should have played a tsume (i.e. c here). But in his own games tsumes don't seem to feature all that much. My hypothesis is that he knew how good tsumes are and so played in such a way that they didn't become an issue. |
|
| Author: | Knotwilg [ Wed Feb 10, 2021 5:21 pm ] |
| Post subject: | Re: Slonection |
To continue on the most popular part of this thread, the decision at move 37: here are the clockwise sequences KataGo sees for each move, hiding the evaluation After the standard sequence, Black A is more attractive than local continuations B, C or D. Black has used the corner aji to get  in and attack the marked stone.We get this well known squeeze sequence. Black connects at A next. White ignores this approach and plays at the bottom After jumping up, White creates aji in the corner and then strengthens the upper left< Standard reduction sequence. Blqck 11 is at A. Next White can attack at B. |
|
| Author: | Bill Spight [ Wed Feb 10, 2021 9:06 pm ] |
| Post subject: | Re: Slonection |
| Author: | gennan [ Thu Feb 11, 2021 1:06 am ] |
| Post subject: | Re: Slonection |
Knotwilg wrote: Following White's marked move I considered A to E (clockwise). What would you play? It looks like a, c and d are the main contestants in other people's replies, while b and e are considered to be too slow. I feel the same, my order of preference being d, c, a, e, b. But in the current AI era, I would definitely consider f and g as well. Didn't you consider those? I might play f, and if white responds in a passive way, continue with g. |
|
| Author: | jlt [ Thu Feb 11, 2021 2:54 am ] |
| Post subject: | Re: Slonection |
So this joseki is still played after Bill's move (Black E18, White D18 and Black D17)? I thought that AI had discarded this variation but apparently I was wrong. |
|
| Author: | Uberdude [ Thu Feb 11, 2021 4:27 am ] |
| Post subject: | Re: Slonection |
jlt wrote: So this joseki is still played after Bill's move (Black E18, White D18 and Black D17)? I thought that AI had discarded this variation but apparently I was wrong. That KataGo thinks it playable is consistent with my pre- and post-AI understanding: locally the wedge is bad for black as white gets a large corner, but if the forcing moves you get on the outside have a sufficiently good direction for future profit, here isolating k16, then it's okay for black. After the hanging connection at a the cut to right of 5 is usually a nice big follow-up for White, but here its value is significantly diminished because doing so 1) will severely blight k16 when black ataris and connects, 2) Black's already fairly stable at c10 so left side is limited. The wedge also works well when k16 is closer at l16 or l17 as then it is more blighted, in fact so well that d18 in reply to e18 is considered a mistake in such positions, something I learnt in a study group with maybe Guo Juan pre AI. Black can do fancy things like attach to it, looking to build a miai to get 2 moves in a row against or save the corner stones. Nice to see KataGo agrees with that. Attachment: Screenshot_20210211_115445.jpg [ 248.43 KiB | Viewed 12703 times ] Attachment: Screenshot_20210211_115511.jpg [ 238.92 KiB | Viewed 12703 times ] |
|
| Author: | Boidhre [ Thu Feb 11, 2021 8:37 am ] |
| Post subject: | Re: Slonection |
Looking at 10k playouts I was seeing three moves that Katago likes a lot after Bill's move and White's response that were all very close in terms of points valuation/playouts: With the idea behind b being that: Similar logic in the other variations. Black can take the joseki at the top and be happy or Black can ask White if they want to answer on the bottom first and either take the joseki or reduce/invade the bottom if they return to the top. E.g. instead if a instead of  on the bottom diagram, Black plays the shoulder hit at b with  from the earlier discussion. I thought these were interesting anyway. E18 as a forcing move before playing on the bottom transitioning into a few different games. |
|
| Author: | Knotwilg [ Thu Feb 11, 2021 11:50 am ] |
| Post subject: | Re: Slonection |
gennan wrote: = It looks like a, c and d are the main contestants in other people's replies, while b and e are considered to be too slow. I feel the same, my order of preference being d, c, a, e, b. But in the current AI era, I would definitely consider f and g as well. Didn't you consider those? I might play f, and if white responds in a passive way, continue with g. I didn't. Move G is met with a top left corner reinforcement. Then Black turns to the bottom again. |
|
| Author: | jlt [ Thu Feb 11, 2021 1:52 pm ] |
| Post subject: | Re: Slonection |
@Knotwilg: there is no  and two  on your diagram. (Edit: the diagram has been corrected.)
|
|
| Author: | SoDesuNe [ Thu Feb 11, 2021 2:24 pm ] |
| Post subject: | Re: Slonection |
Bill Spight wrote: In such positions, this late in the opening, I now ask myself, Where would Shusai play? Before the AI era, I would probably have chosen d, because c invites the jump to g, which looks too good for White. But I don't think that that would have bothered Shusai. So c is now my top choice. I can't speak for Shusai but the c-g-exchange is exactly my reasoning why I don't want to play c. I consider White's gain as more promising than Black's due to Black's corner aji (maybe c ends up being too close to Black's corner position later on?). I would have played d in this position. Funnily enough my first choice in the top left would have been where X marks the spot. But my joseki knowledge is pretty non-existent nowadays so I might miss a continuation where White uses the attachment to make the extension a good move (then again it still faces Black's strong position on the right...). |
|
| Author: | gennan [ Thu Feb 11, 2021 3:35 pm ] |
| Post subject: | Re: Slonection |
Knotwilg wrote: Move G is met with a top left corner reinforcement. Then Black turns to the bottom again. I would not crosscut. I think that is a bit too much. I would be thinking of a counter-hane, but the ladders towards F7 are bad for black, so perhaps the attachment at K4 is unreasonable. If black cannot get more than that, then white is probably happy with those exchanges on the lower side. So maybe black should hold back a bit and play the shoulder hit at J4 instead of K4 after all. |
|
| Author: | gennan [ Fri Feb 12, 2021 4:00 am ] |
| Post subject: | Re: Slonection |
SoDesuNe wrote: Funnily enough my first choice in the top left would have been where X marks the spot. But my joseki knowledge is pretty non-existent nowadays so I might miss a continuation where White uses the attachment to make the extension a good move When black starts with the attachment at D17, white may still revert to the same variation as when black started at E18, but in a different move order |
|
| Page 1 of 2 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |
|
