Life In 19x19 http://www.lifein19x19.com/ |
|
Question on Corner Position (3-3 follow up) http://www.lifein19x19.com/viewtopic.php?f=15&t=17259 |
Page 1 of 1 |
Author: | Kirby [ Tue Feb 11, 2020 7:30 pm ] |
Post subject: | Question on Corner Position (3-3 follow up) |
I was watching one of the games Ryan Li streamed recently, and the following position came about: Ryan decided he wanted to protect in the bottom left here, so he played the marked move. To me, this feels like a "proper" move to play. But I also realized that it's not my first instinct. Namely, if *I* were going to defend the bottom left, I'd probably play here: So why is it worse? Some reasons come to mind: 1. Aji of cutting With my move, there's aji of cutting later: Maybe moves in the marked area may become sente. 2. Follow up? With Ryan's move, the follow up seems decent: That's about all can think of. But I still feel like I'd play the other move in the game sometimes. Because I don't like the feeling when white pushes like this: If it's like the move I'd play with instinct, I feel like white doesn't have this kind of flow, locally: (black's not really pressured) And if like this: Seems a bit easier to play ![]() Also, follow up doesn't seem that terrible with my move: The marked move, or any of the surrounding marked spots seem possible... --- So in conclusion, I'm confused. Ryan's move seems proper - I'm sure it's correct. But the reason is foggy, given the pros and cons I've listed here. What do *you* think? |
Author: | Bill Spight [ Tue Feb 11, 2020 9:03 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
First, Black is going to respond to ![]() ![]() ![]() ![]() ![]() |
Author: | Uberdude [ Tue Feb 11, 2020 9:29 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Oh, I thought the question was going to be why is Better than That's a an interesting little shape study. |
Author: | hyperpape [ Tue Feb 11, 2020 9:52 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Interestingly, if white doesn't tenuki, and instead makes a four stone wall, the solid stretch is relatively common. It's still less used than the diagonal, but not by much. I happened to watch the Shibano-Cho game where Shibano played it. It seems that when this is played, it almost always develops with the immediate push and sacrifice. Not sure what (if anything) that tells you about the original position. |
Author: | Uberdude [ Tue Feb 11, 2020 10:52 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Kirby wrote: Also, follow up doesn't seem that terrible with my move: The marked move, or any of the surrounding marked spots seem possible... The one you chose to illustrate seems pretty bad to me, and a good example of a Kirbyism a.k.a focusing on your opponent's weaknesses and neglecting to give adequate weight to your own. This 2 space jump leaves a glaring weakness at g5. I'm sure you can see that, but you probably think, quite correctly, that if White tries to cut there immediately then you can continue surrounding the wall from d7 area and White's in more trouble then you. But White will (if sensible, maybe the weak players you play don't so encourage and reward your bad habit) take care of the wall with an extension and then later aim at the g5 weak point. You are leaving a debt. If you spend a move to fix it then you just spent an extra move and the jump out was gote instead of sente. This is just considering the local shape of jumping out, in terms of direction it's probably better for black to pincer on the left side and force White to run out whilst black settles on the left to destroy White's potential there. |
Author: | gennan [ Tue Feb 11, 2020 11:59 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Uberdude wrote: Oh, I thought the question was going to be why is [snip] Better than [snip] That's a an interesting little shape study. This question came up in one of In-Seong's lectures or game reviews (perhaps in a Wintercamp?). He said the upper diagram is better shape for black. The lower shape has some defects (white has follow-ups at B3 or G4). |
Author: | Uberdude [ Wed Feb 12, 2020 12:27 am ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Kirby wrote: But I still feel like I'd play the other move in the game sometimes. Because I don't like the feeling when white pushes like this: Sure, these can be nice exchanges (I added black extend to 4 as that's usually better than hane, because hane allows annoying atari for shape and gives white momentum to keep pushing and turn the eyeless wall into eyes and territory) for white to strengthen the wall in sente, but black is also gaining significantly on the lower side with a very thick shape, so it's about equal and depends on position if it's a good idea. But it's important that the white wall here is still without eyeshape or a base, so black can attack it later. For example this left side opening has been quite common in bot self-play: 14 would probably not be played immediately but on right side (e.g. 3-3 invade, or a low approach if there was a 3-4 stone instead), but in terms of the left side shape the follow ups could be as below. If black first then a pincer/invasion at 1 is one idea, it's natural for white to cap to attack but because the wall doesn't have any eyes and black has quite a bit of space on the side and shape points like 7 to aim at if white tries too hard to kill he can die himself. If white first then the jump to e12 is a vital point, simultaneously stopping c12 from running out (though it still has b17 slide) and bolstering the moyo below so that a black invasion like before is one move behind. In that case black usually opts for this gentle reduction, again utilising the f5 weakness (white doesn't get good shape if he pushes out), which also builds his lower side. Note that the left side still isn't quite white territory, apart from b17 slide black still has various annoying moves like e8, which if white answers solidly to confirm the territory black can be content with the forcing exchange, and if white resists then fighting can erupt and black might end up living inside or breaking out with some trades later. Kirby wrote: If it's like the move I'd play with instinct, I feel like white doesn't have this kind of flow, locally: (black's not really pressured) This is true, but also with the kosumi black can tenuki white's push. White's follow up at 3 is simple, if you play a hoping that black submits at b, which can be ok, but you may get a nasty surprise if he ataris at 3 and goes for the trade: the 2 corner stones retain significant aji and the outside ponnuki is very valuable too, so it depends on the board but in general I'd say black prefers ponnuki trade. Also bear in mind that the diagram above white has 1 more move locally, but the position is not as good for him as what the kosumi defence was played to avoid, namely white capturing the outside stone in a ladder: g2 for f4 is a good exchange for black already because it saved f3 in sente. Also a downside of your nobi defence is that white's next moves after next have bigger threats. Returning to Ryan's full board position, imagine it continues like so: white resolves top-right peacefully so I can tenuki after not many moves instead of the big fight they had, and then 3-3 lower right so black gets sente. Now black invades/pincers the left, white jumps out and black extends. Now white 4 is threatening to block at 5, which because of the double-cut weakness the nobi defence left behind would itself threaten to capture the 2 stone, so black is pretty much compelled to answer because allowing white 5 in sente is too painful, and tenukiing white 5 so white can next capture 2 stones is painful too (corner needs a move to live afterwards), and that gives white momentum to keep on running out and helping his group. Nobi defence is not all bad though, as a pro it does have an aim at f5 cut here, though if you let white block at j3 again the double cut weakness bites you in the arse and means you are short of libs to cut at f5, but if black answers with 7 at j3 then white can be happy black answered submissively on the 3rd line and then if f5 does come white might annoyingly dodge with f6 or e7 or so. Compare to with the kosumi, black could merrily tenuki (so white wouldn't play 4) because white has no powerful follow-up (h2 is not small, but it's not sente). |
Author: | Kirby [ Wed Feb 12, 2020 7:27 am ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Excellent discussion! In particular, you made a lot of helpful diagrams, Uberdude. I appreciate it very much. Now, before I totally abandon the church of Kirby, I'd like to drill in on one of the comments here, and how it doesn't totally jibe with one of the passages in the cannon of Kirby: Cannon of Kirby wrote: If they can't punish my weakness immediately, .I might be able to defend in a more efficient way Uberdude wrote: You are leaving a debt. If you spend a move to fix it then you just spent an extra move and the jump out was gote instead of sente. The reasoning above is quite logical - playing ![]() I suspect, in the church of Kirby, ![]() Because this is more consistent with the passage above: he couldn't cut at 'a' immediately, but so I don't have to defend in a slow way like that. Now, if white cuts: I am happy, because I see a target to attack. I feel like black is a move ahead here - I can aim to target white. Coming back to your point, Uberdude - what you're saying is logical: I am leaving a debt. But thinking about things practically, I feel that instead of paying that debt, I have a tendency to try to increase the stakes - maybe my opponent will have a debt of his own. Again, I'm sure there's something wrong with this thought process. But in a weird way, it makes sense to me right now. --- (Off topic) |
Author: | Uberdude [ Wed Feb 12, 2020 8:46 am ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Kirby wrote: Cannon of Kirby wrote: If they can't punish my weakness immediately, .I might be able to defend in a more efficient way That's a fair point, and similar to a favourite proverb/phrase I heard second hand (I think reported from a pro's lecture at an EGC) A wise man once said wrote: If you don't leave weaknesses, you are playing inefficiently. It is such a tough balancing act, not leaving so much weakness your opponent can easily punish you, but not being too defensive and slow. I think one of the lessons of bots, particularly for old Japanese honte style is that the humans didn't leave enough weaknesses (e.g. LZ's dismal view of Kageyama's honte net) and had a tendency to play too conservatively, which can be exploited by forcing them into overconcentration, e.g. the small shimari. Whereas someone like Lee Sedol tends to leave weaknesses in his positions and challenge his opponent to find a profitable way to exploit them against his fierce fighting power. So let me (with a little help from LZ) take up your challenge of how to exploit the g5 weakness if you jump again ![]() ![]() Here's another way to use g5, don't cut, but try to persuade black to connect on dame whilst white does more profitable things. I want black to connect with 17 and then white extends on the bottom whilst attacking the wall on the right, essentially getting j3 for free whilst black pays the g5 debt. Of course black may not be so cooperative, e.g. this 17, but white defends for now and aims to use f4 weakness later. If black aggressively pincers, the double cut weakness wakes up: White getting b4 descent in sente is a very nice benefit, and the black corner still owes a move to live, so once the white group gets out with enough liberties black will need to go back and live and white gets 2 moves in a row outside. If black didn't extend on the top side, but grows the moyo, then again white doesn't want to cut at g5 and start a disadvantageous fight, but use that aim to handicap black's ability to fight strongly, for example white plops a stone in at 8, black caps, and with 0 white is again taking that key point creating juicing follow ups with g5 and e3. If there were no weaknesses there maybe black could play crazy agressive moves a/b and try to kill, but that's not going to work when white comes out the j5 area, so black should probably defend around c so white gets to make a base of sorts at a. |
Author: | Kirby [ Wed Feb 12, 2020 11:25 am ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Uberdude wrote: It is such a tough balancing act, not leaving so much weakness your opponent can easily punish you, but not being too defensive and slow. It certainly is! Since I lean toward leaving weaknesses that my opponent can punish, I'm trying to identify how far I've crossed the line - I don't want to go back too far in the other direction. Thank you for the examples you've given. I'm trying to find a pattern. One thing that sticks out is that in the examples you've given here - you don't cut in a direct and obvious way immediately. You look for a strategy to exploit that weakness, but not directly. That's one reason for me not to leave behind that debt: it lets my opponent come up with plans like this... So stepping back a little bit, why do I like this: compared to something more solid like this (ignoring the fact that the left side is probably a better global choice): I suppose I feel that Diagram A is giving me more influence and control of the board than Diagram B. ![]() ![]() From that perspective, I got some profit from playing ![]() ![]() But I guess the question is, "how much profit?", and perhaps, "how does it compare with the profit white gets by playing on the left and by having options to punish me later, like in the diagrams you've shown?". Quantifying this profit trade off seems useful in terms of identifying the right balance in this balancing act... And I don't really know how to quantify moves like these, except by feeling. I suppose using a bot could help me with this. Identifying the types of variations that could potentially arise, as you've already shown, is also useful. --- So I guess what I can take away from this is that it's tricky to quantify, but I should think more carefully about how my weaknesses can be used, before going for a move that I'm trying to use to be more efficient. Tricky stuff. |
Author: | John Fairbairn [ Thu Feb 13, 2020 3:13 am ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Quote: But I guess the question is, "how much profit?", and perhaps, "how does it compare with the profit white gets by playing on the left and by having options to punish me later, like in the diagrams you've shown?". Quantifying this profit trade off seems useful in terms of identifying the right balance in this balancing act... And I don't really know how to quantify moves like these For decades I have been troubled by the words "territory" and "profit". In particular I have been bothered by the joseki paradigm "profit versus influence", which to me is plain wrong. The problem for me has been that I have no better terms to offer, and so I have just gone with the flow and (mis)used the terms myself. Robert's approach of making a watertight definition of everything just doesn't work, because we are humans, not robots. Indeed, it's worse than that: we are not just humans, we are plain cussed. For example, I have found that any attempt to try to tease out what "thickness" really means has been thwarted because too many people just don't want to listen. It's as if they have built up a rickety apparatus in their minds and they don't want to risk toppling it by adding one more brick to the pile. I think the same sort of thing would probably happen with "profit". The idea of using that word in connection with Black's moves for Diagram A rather took me aback at first. On reflection I could easily see how that came about, but I still found it odd. The reason I have long found "profit" a misnomer is that it implies a gain, and that in turn is such an insidious idea to vain and greedy humans that we end up focusing only on that. The reality is that where we give the opponent a good move, or a good future target, in return it is only a quid pro quo. There is no local profit. I think one way of looking at the Black moves in Diagram A is that they are like erecting scaffolding. If you saw someone erect scaffolding outside a house, you would never say someone has made a profit. You would just say he's made a start, or something like that. The eventual profit (a new roof) is a long way down the road. And, even then, is that really a profit? You have had to pay money to the scaffolders and roofers. It has just been a trade. I don't know hoe best to approach this whole problem, but my instinct tells me that we have to get away from the idea of evaluating a move. We need to evaluate an exchange. In a way we do that when we (wrongly) label a joseki a "profit-influence" exchange. My instinct there is to say we need to think of all the moves there as an exchange of paying for scaffolding and the real evaluation should be what progress that exchange makes to our future plans for a new roof and whether we end up overpaying. But in go all that's hard to articulate, hard to do. At any rate, I don't know how to do it properly even for myself, let alone offer guidance to anyone else on even how to begin. I suppose it's such things that make go interesting, not just as a game but as life lessons. |
Author: | Kirby [ Thu Feb 13, 2020 6:26 am ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
I can see contention with my use of the word “profit”. Maybe it’d be better to say that I feel those moves give “benefit” - for example, in this case, they get more influence than the single one-space jump does. But I suppose “benefit” is not ideal, either, since a one-space jump also has benefits that Diagram A does not, as Uberdude has shown. I like your point about considering exchanges vs. moves, because perhaps that’s part of the problem. There was a Japanese pro that came to the Seattle Go Center a few years back (nickname was “kuma”, or bear, iirc). He compared go to bartering at the marketplace. If I go to the market place and want to get something, I get “benefit” from the product I receive from the transaction, but at a cost (unless the product is totally worthless!). Here, I think the moves in Diagram A “get something” for black, but maybe due to the weaknesses in shape, that benefit is too expensive - it costs black too much. The thing is, sometimes I’m at the marketplace, and nothing has a price tag. So I get stuff that seems like a good deal, but sometimes there are these hidden costs that I miss, so it turns out I actually got ripped off ![]() |
Author: | Bill Spight [ Thu Feb 13, 2020 12:26 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Kirby wrote: Thank you for the examples you've given. I'm trying to find a pattern. One thing that sticks out is that in the examples you've given here - you don't cut in a direct and obvious way immediately. You look for a strategy to exploit that weakness, but not directly. That's one reason for me not to leave behind that debt: it lets my opponent come up with plans like this... So stepping back a little bit, why do I like this: {snip} I suppose I feel that Diagram A is giving me more influence and control of the board than Diagram B. ![]() ![]() From that perspective, I got some profit from playing ![]() ![]() Emphasis mine. Kirby continues in the following quote. John Fairbairn wrote: Kirby wrote: But I guess the question is, "how much profit?", and perhaps, "how does it compare with the profit white gets by playing on the left and by having options to punish me later, like in the diagrams you've shown?". Quantifying this profit trade off seems useful in terms of identifying the right balance in this balancing act... And I don't really know how to quantify moves like these For decades I have been troubled by the words "territory" and "profit". In particular I have been bothered by the joseki paradigm "profit versus influence", which to me is plain wrong. The problem for me has been that I have no better terms to offer, and so I have just gone with the flow and (mis)used the terms myself. I agree about the translation of 実利 as profit. I think that a better metaphor would be hard cash or cash in hand. It is realized territory, or nearly realized territory. Influence, by contrast, is more nebulous: it indicates potential territory, or a reduction in the opponent's potential territory. As for the term, territory, it has a precise meaning at the end of play, unless someone resigns. As we back up from the end of the game the term becomes more and more fuzzy. I don't find that to be problem. In fact, the ability of humans to reason in fuzzy terms is a strength. ![]() John Fairbairn wrote: The idea of using {profit} in connection with Black's moves for Diagram A rather took me aback at first. On reflection I could easily see how that came about, but I still found it odd. Well, it's certainly not 実利. ![]() ![]() John Fairbairn wrote: The reason I have long found "profit" a misnomer is that it implies a gain, and that in turn is such an insidious idea to vain and greedy humans that we end up focusing only on that. The reality is that where we give the opponent a good move, or a good future target, in return it is only a quid pro quo. There is no local profit. While I do not in general like profit as a go term, I can see how it can apply to a quid pro quo. If I have gained more from the exchange than the opponent did, then I may say that I have profited from it. John Fairbairn wrote: I don't know hoe best to approach this whole problem, but my instinct tells me that we have to get away from the idea of evaluating a move. We need to evaluate an exchange. Curiously, I have been thinking along those lines, myself, at least in terms of winrate estimates. ![]() ![]() ![]() |
Author: | Kirby [ Thu Feb 13, 2020 12:35 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Since my idea of "profit" was ignoring the cost of the move, in that I overlooked how my opponent could use my weakness, perhaps I should start saying *revenue* - the amount I get, without considering the cost I paid for it :-p |
Author: | John Fairbairn [ Thu Feb 13, 2020 1:39 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
Quote: Since my idea of "profit" was ignoring the cost of the move, in that I overlooked how my opponent could use my weakness, perhaps I should start saying *revenue* - the amount I get, without considering the cost I paid for it :-p Actually I liked your earlier term 'benefit' for various reasons, one of them I'm sure others will find odd. The word comes from Middle English (Chaucer et al.) and meant a 'good deed'. In other words, the person receiving it (the opponent in go) got the cash-in-hand kind of "profit" whereas the doer (you in go) only got a glow of satisfaction, but perhaps also a note in St Peter's book at the end of the game of life - or go. I think that neatly sums up what uberdude appeared to call a kirbyism. Whatever we call it, all the necessary elements of an "exchange" are implied, with the boot being on the correct foot this time! And there's yet another meta level to all this. Just of we tend to think of "my profit" while forgetting "my loss" (not just in go, of course), we similarly tend to assert our rights while forgetting our obligations (especially nowadays, I might add). Bill: since you mention jitsuri 実利 in connection with "profit" and hard cash, I think it needs to be stressed for others that this is not the normal Japanese word for pecuniary profit. That would be rieki. Jitsuri (the usual go word) has mainly the sense of "tangible advantages", the 利 component also being overloaded with many Confucian nuances, of course, esp. in the still popular Book of Changes (Yi Jing). Influence, as you point out, is the intangible marra. |
Author: | Kirby [ Thu Feb 13, 2020 2:02 pm ] |
Post subject: | Re: Question on Corner Position (3-3 follow up) |
I think the black moves in Diagram A are much better than passes, so are not “losses” in an absolute sense. It’s not just a feeling of satisfaction I have. The stones have positive value. Any loss is from the opportunity cost of not playing a better move. In this case, there’s a weakness I leave for the opponent to potentially exploit. But it doesn’t mean that the stones on the board don’t have value - they’re quite nice to have there, even if the opponent got a better deal. |
Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |