Semeai shortcuts

Something useful for a change, on semeai axioms.

I've been kind of busy recently with the EGF Academy. But I think the time making this post is worth it, as it will collectively save you, my readers, more time in your future games than it took me to make this.
Below I've collected some principles on capturing races.

1.) In the land of the blind, the one-eyed man is king.

 Also known under the less poetic term "me ari me nashi" (have eye don't have eye). It means when someone has an eye vs. a group without eyes, the common liberties belong to the one-eyed group.

By playing the stupid-looking move at 1, Black clearly establishes an advantage in the capturing race of 5 vs. 4 liberties. (Even if White starts playing, White is dead.)

2.) 5, 8, 12, 17

Remember these numbers! A nakade (dead-shape) group with 4 spaces inside has 5 liberties. A nak... actually let me put it in a table:

4 spaces = 5 liberties
5 spaces = 8 liberties
6 spaces = 12 liberties
(7 spaces = 17 liberties, this one is rare and hard to illustrate)

1, 2 and 3 spaces don't have extra liberties other than the visible ones.

These numbers will help you read out semeais faster than ever. Let me explain further though:

Here Black has 8 liberties inside. White has 8 liberties outside.

Playing in the nakade shape counts as taking a liberty. So now Black has 7 liberties inside. White wins by 1 liberty. Feel free to read it out move-by-move if you don't believe me. Then check how much time you could have saved by believing me~

For the slightly more advanced: In a real game, when you count White's territory in this shape, do not forget to subtract 7 (semedori).

Let's look at an example from a real game of mine (I was Black):

I have a nakade 5 shape in the corner. There are 2 stones inside. Then I have the common liberty with White's group. So my group has 8-2+1 = 7 liberties. Using this method, we quickly established my number of liberties.

White's liberties are a simple matter of counting. We can see that White also has 7 liberties, so Black wins by playing first, right?

That's what I thought during the game as well. But looking carefully, White's shape has 2 convoluted cannot-approach-double liberties, so actually White wins by one liberty. That was unfortunate for me.

Fortunately, after capturing Black's group, White's whole-board position is a disaster. Black has sente and controls the whole rest of the board. I also kept the position in the previous diagram (without playing out the semeai), as taking each liberty is an absolute ko-threat. So basically Black can do whatever he wants for the rest of the game, which is what happened. So I won the game after accidentally sacrificing the corner group.
Anyway... :)

3.) In the land of the one-eyed, the bigger-eyed is more king.

In a semeai between 2 groups with 1 eye each, the common liberties belong to the eye with more spaces=bigger eye space.

For White it's actually the same as not having an eye at all. This is the same semeai as in 2.) in spite of White's eye.
(1, 2 and 3 spaces count as the same size of eye.)

4.) Formula for semeai between 2 non-eyed groups

I omitted things like "when both have the same amount of liberties, the one who starts, wins", as I assume you are familiar with that concept. Sometimes, however, there are so many common liberties that you are not sure whether it's a seki or not. There exists a simple formula for that:

Given that: -neither group has an eye (or both groups have the same size of eye),
-it's your turn to play,
-you want to test if you can kill the opponent's group,
you win if

your outside liberties ≥ your opponent's outside liberties + common liberties - 1

Using this formula, we (Black) will try to kill above White's group:

7 ≥ 2 + 6 - 1

This is true, so Black wins by playing first. (If the difference is 1 or more, Black wins anyway.)

To demonstrate it better, White now has one more liberty. If Black tries to kill White:

7 ≥ 3 + 6 - 1

This is false, so Black cannot kill White.

Secondly, can White kill Black?

3 ≥ 7 + 6 - 1

This is obviously extremely false too, so White cannot kill Black either. Therefore the position is a seki.

5.) Semeai that includes ko

-take outside ko first
-take inside ko last

Please try these out by yourselves~

the Go player's posture

When I learned under Kobayashi Chizu sensei, she taughe us about the differences between the habits of European Go players and Japanese conventions (sitting at a Goban on a table).

From what I remember of her teachings, this should be a pretty good example:



Since the guy in the picture is right-handed, his bowl is placed at his right and the lid where prisonders are supposed to be put is in both players' field of vision (though for pros it is impolite to count the prisoners in the lid).
Furthermore you can see that the elbows are off the table, the hand is off the lid until the moment he knows where he would play and reaches for one stone from the bowl which he would place on the board in a confident way.
From my experience it is not that strict at the tournaments, but you should avoid rattling with stones or play with them making a lot of noise because it annoys approximately 100% of Go players. Except of course the one Boom! you make when putting the stone on the board.


Alright! Find 3 mistakes in this picture:

endgame in another nutshell

(cont.)

dedicated to the notorious term double sente.
Double sente means the yose is sente for both players.





What's the size of the endgame in this area?











If white plays first, the sequence is sente for white.




The same goes for black. The difference between this and above diagram is 4 points (marked with the green Xs). We learned that if and endgame is sente for one player and gote for the other, the size is doubled if the gote player gets to play it because the sente player was too stupid to grab it.
As for double sente endgame, both players have the chance to get it for free, so it is hard to calculate the value.* (Someone said that it is worth infinite points.)
However, getting a double sente endgame can give you a psychological advantage. Imagine that your opponent will be all like "oh damn... I should have played that... what if I lose by 2 points now..." and become more overplay-prone.

So much for the theory!





In reality, this kind of endgame is (almost) always more sente for one of the players. White A threatens the black corner (if black blocks once and tenukies, the corner is dead in ko, or alive in ko, whichever you prefer), and black B threatens to ruin white's eastside territory. In a perfect game, whoseever threat is bigger will get the endgame. In a game between two stupid humans, it is the one who is more lucky who will get it.

When you are unsure whether your opponent might tenuki your supposedly sente move (this goes for single sente moves too), there is this rule of thumb (found on senseis):

If you hope your opponent doesn't answer your move, your move is probably sente.



But keep in mind that there are situations where playing endgame too early can be bad.




Feel the power of white's moyo!






And now look how black's moves have increased white's power (in sente).




*The nerdy way to calculate is to assume that everything is gote and calculate the net value of the move and add the half size of the continuation to a gross value, and do that for both sides. However, it is probably easier to do an overall score estimation for every move than calculating that (and see rule of thumb).

endgame thinking in a nutshell

(for the clueless endgame newbies)


Endgame, also referred to as yose, is the end of the game, as the name might suggest. In Go, the value of moves get smaller towards the end of the game. And as the game approaches the end, it becomes possible to express the value of a move in an exact number points.



what's the value of investing a move here?











to count the value of an endgame move, we compare the situation as black plays there and the situation as white plays there. Here black has 6 points and white has 5 points.







and if white plays first, black has 5 points and white has 6.
The endgame is worth 2 points (and for the more advanced, there is also a difference of 3 ko threats).



Since the endgame is gote for both sides, when you try to estimate the score you can add the two triangled stones. It means that both players have an equal chance of getting the endgame. It is pretty accurate, if you want to be more accurate, you'll have to play the endgame to the end in your head.






Next is this situation... Try to take a guess about the value of the yose first.







This situation is different from the previous one since white has to protect at 4. The sequence is sente for black.





If 4 is omitted, bad things happen.


So we learned we have to compare the situation when black plays first to the situation when white plays first. Note that for white, the sequence is gote. The difference in territory is marked by the 3 green Xs. For black, the endgame is sente 3 points. For white it is gote 3 points, which is called reverse sente since it prevents black from playing it in sente. In this case the value is doubled (an approximation for your brain), 3 points reverse sente = 6 points gote.



For score estimation, it is safer to assume that black gets to play his sente sequence.

to be continued