|time||9 minutes plus 10/30s|
The first round started at 16:00 UTC.
|1||Zen19S||B1474 W0112 B112R W0206||W022 B17R W110R||B082 W192 J15 W11736||W15R B113R W119R||W13R B114R||W16R B116R||B11R W11838||15½||208½||139||Winner|
|2||CrazyStone||W0474 B1112 W012R B1206||B035 W115R B117R W118R||B112 W0774 B1194||B16R W19R B114R||W12R B016R||B15R W110R||B188 W11316||15||211||146½|
|3||AyaMC||B122 W07R B010R||W135 B015R W017R B018R||B05R J11 B1122||B11R W18R||W14R B113R W120R||B19R W114R B119R||B164 W11646||13½||203||101½|
|4||DolBaram||W182 B092 J15 B01736||W012 B1774 W0194||W15R J11 W0122||B032 W116R B120R||B16R W110R B118R||B14R W113R||W1244 B114R||13||210||97|
|5||HiraBot||B05R W013R B019R||W06R B09R W014R||W01R B08R||W132 B016R W020R||B17R W111R B115R||W12R B112R W118R||B144 W1104 B1172||10||196½||52|
|6||NiceGo19N||B03R W014R||B02R W116R||B04R W013R B020R||W06R B010R W018R||W07R B011R W015R||W11R B08R W117R||B1510 W192 B11212 W11910||7||189½||29|
|7||MCark2||B06R W016R||W05R B010R||W09R B014R W019R||W04R B013R||B02R W012R B018R||B01R W18R B017R||B03R W17T B11112 W1158 B12010||5||183½||11|
|8||gnugo3pt8||W01R B01838||W088 B01316||W064 B01646||B0244 W014R||W044 B0104 W0172||W0510 B092 W01212 B01910||W13R B07T W01112 B0158 W02010||1||197||5|
Black won 38 games, White won 40 games, and there were 2 jigoes.
Seven players entered. As last month NiceGo19N, Detlef Schmicker's latest version of oakfoam, was using a Convolutional Neural Network.
To make the numbers even and avoid byes, I included gnugo3pt8.
|AyaMC vs CrazyStone|
|After both players passed|
In round 3, the two undefeated players AyaMC and CrazyStone played. They both passed in the position shown to the right. They were both right to pass: the game is over, the top right is seki, and AyaMC (white) is four points ahead. However, they disagreed about the status of the seki. Play resumed, two more stones were played (white ones, at c8 and j5), and both players passed again. This time they agreed on the status of the seki.
It is some time since I have noticed the "clean-up" procedure being used. While it was disappointing that one of two such strong players failed to understand a relatively simple seki, it was pleasing that both managed the clean-up procedure and got the right result.
|NiceGo19N vs HiraBot|
|After Black 89 (pass)|
In round 7, Black, Hirabot, passed in the position shown to the left, and NiceGo19N resigned. If it had also passed, the result would have been jigo. But NiceGo19N was using CNN, and did not understand that pass was possible. If White is not allowed to pass, he will soon lose his only group.
After this game, NiceGo19N's operator disabled its CNN code, which in any case had been trained on only 137 games.
CrazyStone was unlucky not to win the tournament, coming half a point behind Zen19S. It had
two wins and two losses against Zen19S, and achieved slightly higher win ratios than Zen19S
against third-placed AyaMC and fourth-placed DolBaram. But its surprise loss to sixth-placed
NiceGo19N cost it the first place in the tournament. I look forward to seeing more games between
Crazy Stone and Zen – and between all the strong programs that competed in this event.
Players receive points for the 2015 Annual KGS Bot Championship as follows: