Here's a chart from ps.waltheri showing the rapid rise in popularity of this shape post AI. Not unheard of before though.
Attachment:
Screen Shot 2022-01-15 at 12.48.48.png [ 23.49 KiB | Viewed 4877 times ]
| Life In 19x19 http://www.lifein19x19.com/ |
|
| Good shape? http://www.lifein19x19.com/viewtopic.php?f=15&t=18570 |
Page 1 of 1 |
| Author: | Wodan58 [ Sat Jan 15, 2022 1:00 am ] |
| Post subject: | Good shape? |
Is this good shape? |
|
| Author: | John Fairbairn [ Sat Jan 15, 2022 3:57 am ] |
| Post subject: | Re: Good shape? |
Wrong question. The right question is: is there a better shape? Yes, usually. Depends on the wider position. |
|
| Author: | Uberdude [ Sat Jan 15, 2022 4:59 am ] |
| Post subject: | Re: Good shape? |
The pre-AI standard theory I learnt was that this shape was generally to be avoided as, whilst the 3-3 invasion is in principle killable, in practice it's messy with all sorts of bad aji and problems outside. So for that reason the enclosure with 1 knight's move and 1 diagonal (move f17 to e17) was favoured, as then it's easy and clean to kill the 3-3. But nowadays AI seems to prefer this shape over the tighter one which is a tad inefficient and AI can deal with the complexities of this one (or indeed AI also likes 1 small knight and 1 big knight). There's also things like the attaches here: http://ps.waltheri.net/database/game/79223/. That's not to say the tighter enclosure still isn't a better idea for us weaky humans, as it's easier to handle (e.g. how do you count this corner, is it your territory?). I do play this one now though to have fun and learn about new things. |
|
| Author: | Uberdude [ Sat Jan 15, 2022 5:49 am ] |
| Post subject: | Re: Good shape? |
Here's a chart from ps.waltheri showing the rapid rise in popularity of this shape post AI. Not unheard of before though. Attachment: Screen Shot 2022-01-15 at 12.48.48.png [ 23.49 KiB | Viewed 4877 times ] |
|
| Author: | kvasir [ Mon Jan 31, 2022 2:29 pm ] |
| Post subject: | Re: Good shape? |
Uberdude wrote: Here's a chart from ps.waltheri showing the rapid rise in popularity of this shape post AI. Not unheard of before though. There is something wrong with these graphs on ps.waltheri. I have seen "popularity" higher than 900% in 2018 for a simple hoshi. Maybe everything from 2018 and later is put at 2018 in the graphs, not sure how it gets to such high numbers unless a game with 4 hosihs counts as 32 occurrences (rotations and mirroing). I'd be careful to draw conclusions, the wrong graphs aside, because it is not only the frequency of the double enclosure positions that matters but also the frequency of playing the small knights move when there is nothing around. The increase in the occurrence of the small knights move when there is nothing close by is also large. My rather rough estimate (just based on the rather defected information from ps.waltheri) is that playing the double enclosure is 5 times more common than before, going from 1.8% to 8.4% of cases that see black playing the third move in the corner. The original small knights enclosure is 2 times more popular so we see more of this but it is also also that almost 15% of the games on ps.waltheri are from 2015 and later. It would appear it is still not that usual but then there is the dilemma of how to draw the search area and the fact that a lot more games are freely available in recent years. There is a huge bias in selecting the games, both in what is recorded and published in general and in what appears in ps.waltheri specifically. One example is that a search on an empty board only returns Chinese games from March 2020 at the aptly named event "not available". Then there is the fact that it is just 99 games or so in total, if you use a wide search area. |
|
| Author: | John Fairbairn [ Tue Feb 01, 2022 3:40 am ] |
| Post subject: | Re: Good shape? |
If I may be allowed to go back to my point above: Is this good shape? Wrong question. The right question is: is there a better shape? Yes, usually. Depends on the wider position. What the question triggered in me was a reference to this sort of position in Game 1 of the Kamakura match. It was White (Go) to play in the position below. Go played A. Go showed he was just as aware of pretty shapes and their names as any amateur, because he said that if he had played B he would have a position like the famous swordsman Miyamoto Musashi - a sword in both hands. But that was an entirely nugatory consideration. He declined B because it left the corner open to invasion at D. The tsume at C was rejected because the invasion at D would then be even worse. But A was chosen not just for negative reasons. Even locally it had the merit of a nice follow-up move at C. Its real merit, however, was that in terms if the overall game he was most concerned with building up his moyo on the left side and didn't want to have any distractions on the right side. Choosing a move because it has pretty shape is like seeing a nice sofa in a shop and buying it because it is so nice. But once it is delivered to your home, you may find it doesn't fit the way you hoped: you face the window where the sun streams in, the tv is in a place that makes you crook your neck, or you have to shift other furniture into awkward and ugly positions. The sofa still looks nice, but the exercise has been a disaster. The overall position and requirements have been ignored. An easy way to spot if you are falling into this kind of predicament is to ask whether you can substitute "good shape" with "pretty shape". If you can, alarm bells should ring. On a different tack, I also agree with kvasir: the graph itself is a pretty shape, not a good shape. Apart from the reasons he adduces, I would argue that it is highly unlikely that any pro would play a move just because it was a pretty shape. If he did play such a move, it would be for extraneous reasons elsewhere on the board, such as those outlined by Go above. So, if there is a genuine cluster effect as shown by the graph, it may be because there is a cluster of extraneous factors that just happen to make the pretty move the best in these circumstances. In sofa terms, many rooms are rectangular. Let's say a three-seater sofa will fit on the short sides, a four-better sofa on the long sides. Let's also say a new tv has become fashionable that will cover the entire wall on one of the short sides. Sales of three-seater sofas start rocketing, four seaters go into decline. That surely has nothing to do with three seats being better than four or vice versa. On yet another different level, it is always useful to go back to the original source. "Good shape" is the English rendering of the Japanese katachi, which actually just means "shape." But the Japanese term has two distinct uses. The distinction is easily spotted. It tends to gets lost in English because we add 'good.' In the diagram below, the triangled move by Black is referred to as katachi and in that case is defined as "vital point for prophylactic defence" (mamori no kyuusho). In commentaries the deictic THE is typically implied. It may be THE only move. It is the type of move exemplified also by Go's move A above. It is local, but because it is prophylactic it has deeper strategic import. On the right side we see the other Japanese usage. The idea there is one of efficiency, with no real mamori overtones. You are usually defending to a direct stimulus (i.e. ukeru) rather than prophylactically (i.e.. mamoru). It is much more local than strategic. Here, the point is that the triangled White move is better 'shape' locally than a move at A. If we look at the OP position in the light of the Japanese usages, we see that it is neither fish nor fowl. We need extra information before we can make an assessment. We need to know what sort of room we want to put our sofa in. |
|
| Author: | Knotwilg [ Tue Feb 01, 2022 5:59 am ] |
| Post subject: | Re: Good shape? |
I think it is well understood that "good shape" should not be reduced to be "beautiful shape" or any superficial approach to it. Something is "good shape" if the stones accomplish their goal in an efficient manner. That said, the concept would not be useful if it's simply a synonym for "good move". The introduction of the visual connotation of "shape" means that you get an inspiration or confirmation of a move being good, because you recognize the shape. There must be some kind of reduction in order for the concept to be useful. It's similar to "joseki". You recognize an established pattern, so you know this is a candidate for the best next move or sequence, without guarantee and with the risk of ignoring non standard moves. A bamboo joint and a table shape are well known "good shapes" for connection, while the empty triangle is a well known "bad shape". Through my AI fueled analysis I have come to suspect the table shape as an overestimated idea, in the sense that I play it too often when it's not efficient and rather slow. In this forum we have discussed the empty triangle even recently. On the OP, the answer needs a global context. Uberdude has IMO given the perfect answer. |
|
| Author: | Wodan58 [ Wed Feb 02, 2022 2:49 am ] |
| Post subject: | Re: Good shape? |
Global context: The shape was mentioned by Andrii in a game he commented, though not played. The shape was played in the following two games: First game: Ali - Oscar Second game: Pavol - Ilya These games were played in the 2021 Grand Prix Final; in both games the latest move was played on the left side of the board. |
|
| Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |
|