|Life In 19x19
|Underappreciated AlphaGo contribution: white 6 shimari
|Page 1 of 1|
|Author:||Uberdude [ Mon Feb 22, 2021 1:25 pm ]|
|Post subject:||Underappreciated AlphaGo contribution: white 6 shimari|
I was prompted to make this post here due to reddit's lack of diagrams, but I think one of AlphaGo's most important contributions to modern opening theory is often overlooked with all the direct 3-3s nowadays, and that's how patient white can be (with the 7.5 Chinese komi). Exemplified by this white shimari for move 6 often seen in the Master games, rather than the prior human moves of splitting (a) or approach(b more active, more popular in 2010s) to immediately break up black's side.
|Author:||dhu163 [ Thu Apr 08, 2021 4:26 pm ]|
|Post subject:||Re: Underappreciated AlphaGo contribution: white 6 shimari|
Even much later in the game, I think white playing a or b is very rare. It is normally 3-3, tenuki and let black defend, attach 4-4 in upper left, or approach from outside.
Tentatively, a principle to sum this up is that a single weak white stone at the top actually helps B make use of the shimari and influence. A single stone fails to stop B making an extension from the shimari or corner, while remaining weak.
Hmm, the below turned into a long winded essay and I'm not sure how much sense it actually makes. I still wonder if such things can be quantified and/or made more rigorous, or shown to not be the right way to think about things. What do you think?
To exaggerate a bit overly in favour of this principle:
a seems to be traditionally called a "wedge", probably based on Japanese. In Chinese it is called a 分头, literally dividing ends (or heads). This invokes the powerful feeling of wedge tesujis that hurt the opponent's shape. But from the being only move pre90s, it faded from top professional Go as b became the most popular. AI seems to regard a as a junk move, premature at best.
With c next, this was extremely popular back in the day leading to a fuseki that AI says is terrible for W.
a has the feeling of 'middle of nowhere'. It is normally only played by AI when the opponent is very solid all over the board and hence the opponent can quickly convert a moyo on the upper side into territory (or a rapidly expanding moyo) if you don't play there immediately. And normally it is the upper right corner that gets particularly solid from fighting on the right, in which case the vital point is likely to be one to the left of a.
Where else is there to play other than around a? Even here it might be too passive, failing to attack black and too slow to get into the centre when in the opponent's moyo. Note that if the upper left shimari was at all different, leaning on the shimari might be more appropriate. The one space shimari is a solid connection, doing more to split the upper side from the left side.
For example, after a 3-3 invasion in the upper right, if black blocks on the top for a wall but tenukis, then a becomes a perfect move to make a living group blocking a good deal of what would otherwise be black's giant moyo. The other 5 moves around it are very interesting too. In the ideal scenario, black has nirensei on the upper side, white invades 3-3 in both in sente, black blocking for the top and white plays on the 4th line, one below a, and still has a lot of empty space on both sides to make life or maybe even territory.
Is white dreaming? c is next
b is not going to be too bad in the eyes of AI as it is active.
b has the feeling of 'attacking the opponent in their own area with a weak stone'. Based on the position, black chooses to defend on the right or pincer at the top. The whole idea of sacrificing an approach stone to get the corner is perhaps slightly flawed at the AI level (I first saw a tewari when Tiger showed me) at least in this setup. The advantage of invading 3-3 directly is that at least you are attacking with a strong stone and you short several liberties of the 4-4 stone. A 3-3 also has the feeling of "on the outside" as it has access to the right. So it is often a reduction that threatens to build on the right side. b threatens to nullify the upper side and build where the opponent had potential. It is this combination that makes b attractive, but it is difficult to get the upper hand when both sides are strong.
c is next. In the upper right, black's stones look like ajikeshi moves too close to w's corner. They leave some weaknesses in w's shape, but they are hard to exploit when black is low on liberties and w is threatening such a large area.
Approaching from the outside instead also has the flavour of taking what the opponent had and making it yours but more softly, with follow ups that are big because they are next to empty space which seems more reasonable. The follow ups that are generated by b emphasise either access to the centre via access to the upper side or an attack on the upper right corner. But there is always the worry that black can extend from the shimari while attacking white.
Follow ups generated by a are only worse for territory, but a does prevent black building a large area that develops into the centre. A problem with preventing such a plan of the opponent is that you could have reduced such a moyo from the centre anyway.
If you have access to the sides, all you need to worry about is the centre. is probably premature and too deep, but it is still awkward for B. Sometimes is the best move.
Another thing is that if you increase aggression on Katago, it plays more of this sort of move and less on the upper side to invade a moyo (which is the opposite of what I would do in a handicap game pre KG). The aggressive KG, normally W plays 3-4, shimari, extension from its side and avoids invading the opponent's area.
|Page 1 of 1||All times are UTC - 8 hours [ DST ]|
|Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group