Life In 19x19

Hi! I'm an avid board gamer that has finally learned to appreciate abstracts and decided to learn go once and for all. So far I'm finding it very rewarding and awesome.

So today I played a match against the Aya windows app running Aya 6.34 (I believe not MC). I lost, as usual... but this time I was confused.



Aya


Now, by my count I have the following:

Me: Territory(88) + Captures( 7) = 95
Aya: Territory(26) + Captures(25) + Komi(6.5) = 57.5
Which means Black (me) won by 37.5, which is very different from Aya's claim that white won by 42.5.

So, I fed the SGF to a couple of other apps.

qgo

Me: Territory(88) + Captures( 7) = 95
Aya: Territory(26) + Captures(25) + Komi(6.5) = 57.5
Yay! This matches with me! Let me try another one!

Gobandroid
Me: Territory(88) + Captures( 7) = 95
Aya: Territory(17) + Captures(25) + Komi(6.5) = 48.5
Black wins, but by 46.5 and not 37.5...

Pandanet(Go)
Me: Territory(86) + Captures(10) = 96
Aya: Territory(26) + Captures(25) + Komi(6.5) = 57.5
Black wins, but by 38.5, which is pretty close to my estimate, but different territory and captures!
If I remove all of the stones which I think are dead stones then the score is 106.5 for white and 99 for black, having white win by 7.5. *shrug*

GridMaster Free
Me: Territory(88) + Captures( 7) = 95
Aya: Territory(23) + Captures(25) + Komi(6.5) = 54.5
So, once again, black wins but this time with 40.5


So. Everybody except Aya agrees that Black won although there are some differences in the actual score. So, is Aya right and everybody else wrong ? If Aya was wrong, was my counting correct?

HELP!

White won by 39.5 points. For some reason, aya doesn't count the three points along the right side of the board @ T12, T13, and T14.

The reason you see difference in score is because of dead groups on the board. Some go programs cannot detect which groups are alive and which groups are dead, so as the user, you must click on those groups to remove them from the board.

Dead black groups count as if white captured those groups. Dead white groups count as though black captured those groups.

In the final board position, stone groups at:
* A16
* B19
* D18, C18, E18, F18, F17
* H18
* G15
* K10
* K6
* L5
* L2, M2
* S2
* C2, C1
* E1,E2,F2,F3,G2,G1
* R6

are all dead.

When you finish the game in your program, if the program cannot detect the dead groups, you have to click on them so that they count as the opponent's territory.

For example, in your screenshot from qGo, these dead black stones aren't marked as dead, so the scoring isn't complete, yet. You must mark them as dead to get the correct score.

* Note, if these groups had two eyes and were alive, they could stay on the board and not be dead. Unfortunately, this is not the case. For example, F2 has only one eye, with no room to make a second eye. Therefore, it is a dead group. White could play stones to remove it from the board, but it is not necessary since it is impossible for black to avoid being captured.

So then that's two points for each dead stone. That brings white's score to 103.5 and black's to 97 which is only a difference of 6.5 and not 42.5.... I must be missing something

Nevermind, I figured it out!

For the most part, at least. I still only get 136.5 for white so there's 3 points somewhere that I'm not seeing, but I realize my original errors now. Thank you!

I get a White win by 39.5.

Edit: I see that Kirby and I agree.



Your problem was not recognizing your dead stones. Most scoring programs rely upon humans to mark dead stones. That is why so many of them scored this game incorrectly.

Kirby wrote:White won by 39.5 points.

Bill Spight wrote:I get a White win by 39.5.

Oh good! At least that now matches up with my corrected scoring.

Kirby wrote: * Note, if these groups had two eyes and were alive, they could stay on the board and not be dead.

Heh. I know it doesn't look like it, but I was aiming for two eyes with most of those groups. I was only deliberately successful with the group in F10.

Bill Spight wrote: Your problem was not recognizing your dead stones. Most scoring programs rely upon humans to mark dead stones. That is why so many of them scored this game incorrectly.

Indeed. At some point I knew this but I think I have been relying on the apps for scoring so much that by the time I actually decided to score it myself I was all thrown off by all of these apps giving me different answers even when I knocked out dead stones when using Pandanet(go).

I really appreciate you guys helping me work through this!

Hi Hokie,

Welcome to Go.

good.

did you considering blocking at K17 ?
( When pushed at K17, you still replied. )

good. Now imagine if you had simply blocked at K17 on -- compare that result to this current result.

good.

very strange (a few reasons). Locally, why would you not play at K18 instead ? ( Globally, there are better moves. )

first big mistake. W would push through at D5, crushing your shape. You should block at D5, just like .

W should push through at D5 and destroy your local shape.

EdLee wrote:Hi Hokie,

Welcome to Go.

good.

did you considering blocking at K17 ?
( When pushed at K17, you still replied. )

good. Now imagine if you had simply blocked at K17 on -- compare that result to this current result.

good.

very strange (a few reasons). Locally, why would you not play at K18 instead ? ( Globally, there are better moves. )

first big mistake. W would push through at D5, crushing your shape. You should block at D5, just like .

W should push through at D5 and destroy your local shape.

Oh snap! Thanks for reviewing it!

No, I didn't consider K17. I was very focused (for some reason) on a knight's move towards the inside. One thing I'm noticing about how I've been playing these couple of weeks is that I seem to be hesitant to reach the edges. I don't understand how that works in my head as it is obvious that territory is easier to acquire by using the edges. Still, my thinking was "extend to the middle!"

I see what you mean! That stone in J14 is doing nothing for me right now when it could have been making that shape stronger.

I think I was thinking here that this might help that shape by "blocking" white from coming around on the stone in L17 and possibly help create some eyes. I was just now about to figure out my reasons for not placing on K18 which forced me to look at that location much more clearly and, man, that would have been considerably smarter. Why on earth did I chose to leave it wide open for W?! Either way, you're right. It wasn't a good global move. Overall a wasted move.

Interesting. I would have never considered this a huge mistake. But now that you're pointing it out, it's a very silly move indeed! D5 would have been a much better move for both me and as you point out next. I'm not really doing much on E6.

Thanks for the input and the welcome. That was fun!

it is obvious that territory is easier to acquire by using the edges. Still, my thinking was "extend to the middle!"

Yea, in many cases, the corners first, next the sides, and finally the center. But there are infinite exceptions.

I would have never considered this a huge mistake.

A very important class of shapes to start to learn. Please see also toothpaste.