Hundred and eighteenth KGS Computer Go Tournament

Sunday December 6th, 2015

These results also appear on an official KGS page.


format18-round Swiss
board size13×13
time9 minutes plus 10/30s


The first round started at 16:00 UTC.

Result table

abaku AyaMC WeakB ManyF Raybo MCark NiceG matil
B13R W010R B112R W117R W14R B19R W111R B15R W113R B116 B02R W17R B118R B18R W115R B11T W114R W16F 16170148Winner
2AyaMC W03R B110R W012R B017R
W02R B08R W114R B118R W1616½ B17 B19R W115R B11R W111R B14T W116R W15T 13174102
3WeakBot B04R W09R B011R B12R W18R B014R W018R
B11T W110R B017R B13R W11219½ B15R W116R W16R B113R B17F 1217388
4ManyFaces1 W05R B013R W016 B0616½ W07 W01T B010R W117R
B14R W18R B114R W13R B112R B19R W115R B118R W12T 1116466
5Raybot W12R B07R W018R W09R B015R W03R B01219½ W04R B08R W014R
W16R B113R W117R B15R W111R W1137½ 915944
6MCark W08R B015R W01R B011R W05R B016R B03R W012R B06R W013R B017R
W12T B07R W110R W14F 614610
7NiceGo19N W01T B014R W04T B016R B06R W013R W09R B015R W018R W05R B011R B02T W17R B010R
W13F 51516
8matilda B06F B05T W07F B02T B0137½ B04F B03F
0720Quit before round 8

Excluding forfeited games, Black won 26 games and White won 31.


Eight players registered. These included matilda, by Gonšalo Mendes Ferreira, a newcomer to these events (it had hoped to play in the November tournament, but had to withdraw).


In round 1, ManyFaces1 had trouble connecting. It had not played on KGS since July, and since then there had been new releases of both kgsGtp and Java. It lost its game, on time, to WeakNot, without being able to make any move.

Also in round 1, NiceGo19N and abakus failed to agree on the status of some stones after both players had passed. The game entered the clean-up phase, and NiceGo19N failed to make a move, and so lost on time. In fact it was a won position for abakus.

In round 2, ManyFaces1 connected successfully for its game with matilda. But after making six moves matilda vanished, and lost on time.

Also in round 2, raybot beat abakus.

Before round 4, matilda reconnected, and started to play AyaMC. But it disconnected again, and lost on time.

In round 7, matilda withdrew from the rest of the tournament.

WeakBot vs Raybot
At the game end

In round 12, Raybot appeared to understand that "bent four in the corner" is dead, and when it and its opponent WeakBot passed, in the position shown to the right, Raybot claimed that WeakBot's dead corner was indeed dead. WeakBot disputed this, and the game went into the clean-up phase. Raybot did not try to capture the bent-four group, nor the single dead white stone, but passed; so WeakBot was credited with the win. I hope that Raybot will implement the clean-up phase sometime.

Annual points

Players receive points for the 2015 Annual KGS Bot Championship as follows:

Many Faces of Go2

Details of processor numbers, power, etc.

abakus, running on 8 nodes each with two Intel Xeon E5-2670, 2.6GHz and 64 GByte main memory (128 cores total) and each with a Tesla K20 (8 GPUs total).
Aya, MC version, running on i7 980X 3.3GHz 6 cores.
Many Faces of Go, running in the Amazon cloud, on an 18-core Intel E5-2666 v3.
matilda running on 6 core i7 @ 3.2 GHz and 32 GiB RAM
MC_ark, running on a Xeon E5-2450 2.10GHz 8-core HT on*2cpu + GPU(G200eR2).
oakfoam, running on an i7-4790K / GTX-970.
Ray running on a dual Xeon E5-2687W v2 (total 16 cores) with 64GB memory.
HiraBot running on an i7-4790k 4 cores 8 threads. HiraBot use neural network data. 18 core Intel E5-2666 v3.