Tsumego concepts
Posted: Sun Feb 13, 2022 4:55 pm
Among other things, the thread
viewtopic.php?f=10&t=18569
has encouraged me to think about concepts on tsumego, one of the most fuzzy and difficult parts of Go to play correctly.
The below is a draft of some ideas. It might not make that much sense.
There are general proverbs sich as hane, vital point, cut. And I tended to just memorise shapes to know their status and correct move.
Probably there isn't much better, but there should still be a list of some tactics from level 0 onwards, even if we know the full list is infinite. Tactical reading tsumego is probably the area in Go most like Chess.
I may add diagrams later.
Board positions (neighbours and type based)
0. Combinations of the below where one type is a follow up to another type.
1. An entrance point where both sides have solid stones adjacent. In general, a cutting/connection point for both sides. (The attacker can connect to defender's eyespace thereby breaking it, the defender can connect two boundary areas)
2. A cutting move that isolates at least one chain, thereby ensuring that other spaces adjacent to it cannot become an eye.
3. A cutting move that reduces the liberties of all adjacent opponent chains (both current and potential future chains).
4. A ponnuki (or other capturing shapes) isn't connected without the centre point and it takes several more moves to turn it into a solid eye. If the ponnuki must be connected, that costs a liberty.
5. Atari seal and variants are when a group is almost connected to another group (or just the outside) but only via one stone in a bottleneck. The opponent can cut that stone from the group even if the cutting stone dies and use the shortage of liberties to atari block the bottleneck from the other group. Due to 4, the bottleneck remains unconnected, and though it is ko, the opponent capture the bottleneck tends to still be a big move, making it a heavy ko.
6. Diagonal moves tend to help both notches at once. The centre of a farmers hat or cross five tends to be very big even if not all the arms are complete yet since it sets up multiple options. Claiming a bottleneck allows conquer and divide. When attacking, even if you can't save the bottleneck stone, you might still get seki.
7. Instead of playing solidly, adjacent to your support, if you have two close enough supports, it may be more severe to jump in the middle with a miai connection to each support.
8. If your opponent has a weak chain, you have additional influence over its liberties since they eventually threaten to capture the group, giving you more liberties. These can act as supports to allow you to play more deeply.
Strategy (goal based)
0. If your opponent is threatening your weak point with a weak group, consider attacking their weak group before defending.
Semeai (maxims)
0. Block your opponent from playing where there are more liberties and strengthen your weaker surrounding stones by driving them against your stronger groups until they have nowhere to run (or you have gained enough from the forcing). Attacking with diagonal moves, one space jumps or knight's moves is often the fastest way to seal and the cutting points won't be an immediate problem if your opponent's group is weaker than your own. If you have enough liberties and local support, elephant jumps and larger loose nets may be playable.
--
Can the complexity of a go problem be measured in terms of such concepts rather than just the raw variation tree?
viewtopic.php?f=10&t=18569
has encouraged me to think about concepts on tsumego, one of the most fuzzy and difficult parts of Go to play correctly.
The below is a draft of some ideas. It might not make that much sense.
There are general proverbs sich as hane, vital point, cut. And I tended to just memorise shapes to know their status and correct move.
Probably there isn't much better, but there should still be a list of some tactics from level 0 onwards, even if we know the full list is infinite. Tactical reading tsumego is probably the area in Go most like Chess.
I may add diagrams later.
Board positions (neighbours and type based)
0. Combinations of the below where one type is a follow up to another type.
1. An entrance point where both sides have solid stones adjacent. In general, a cutting/connection point for both sides. (The attacker can connect to defender's eyespace thereby breaking it, the defender can connect two boundary areas)
2. A cutting move that isolates at least one chain, thereby ensuring that other spaces adjacent to it cannot become an eye.
3. A cutting move that reduces the liberties of all adjacent opponent chains (both current and potential future chains).
4. A ponnuki (or other capturing shapes) isn't connected without the centre point and it takes several more moves to turn it into a solid eye. If the ponnuki must be connected, that costs a liberty.
5. Atari seal and variants are when a group is almost connected to another group (or just the outside) but only via one stone in a bottleneck. The opponent can cut that stone from the group even if the cutting stone dies and use the shortage of liberties to atari block the bottleneck from the other group. Due to 4, the bottleneck remains unconnected, and though it is ko, the opponent capture the bottleneck tends to still be a big move, making it a heavy ko.
6. Diagonal moves tend to help both notches at once. The centre of a farmers hat or cross five tends to be very big even if not all the arms are complete yet since it sets up multiple options. Claiming a bottleneck allows conquer and divide. When attacking, even if you can't save the bottleneck stone, you might still get seki.
7. Instead of playing solidly, adjacent to your support, if you have two close enough supports, it may be more severe to jump in the middle with a miai connection to each support.
8. If your opponent has a weak chain, you have additional influence over its liberties since they eventually threaten to capture the group, giving you more liberties. These can act as supports to allow you to play more deeply.
Strategy (goal based)
0. If your opponent is threatening your weak point with a weak group, consider attacking their weak group before defending.
Semeai (maxims)
0. Block your opponent from playing where there are more liberties and strengthen your weaker surrounding stones by driving them against your stronger groups until they have nowhere to run (or you have gained enough from the forcing). Attacking with diagonal moves, one space jumps or knight's moves is often the fastest way to seal and the cutting points won't be an immediate problem if your opponent's group is weaker than your own. If you have enough liberties and local support, elephant jumps and larger loose nets may be playable.
--
Can the complexity of a go problem be measured in terms of such concepts rather than just the raw variation tree?