Spring 2013 Slow KGS Computer Go Tournament

March 3rd - 6th, 2013

These results also appear on an official KGS page.


format12-round Swiss
board size19×19
timeThree hours each, sudden death


The first round started at 22:00 UTC on March 3rd

Results table

Crazy Zen19 AyaMC pachi nomiB ManyF fuego NiceG MCark Orego
B19R W110½ B14R W18R B12R W111R B13R W112R B15R W17R W16R W11R 127878Winner
2Zen19S W09R B010½
B16R W112R B03 W01R B17R B12R B18R B111R W15R B14R 87840
3AyaMC W04R B08R W06R B012R
B15R W19R W11½ B111 W13R W12R W17R W110R 87838
4pachi W02R B011R W13 W05R
B14R W110R B06R W112 W19R B17R W18R W11R 87638
5nomiBot W03R B012R B11R W07R B09R W04R B010R
W18R W12R B15R B111R B16R 68226
6ManyFaces1 W05R W02R B01½ W011 W16R B012 B08R
B14R W110R B19R W13R W17R 67525
7fuego19 B07R W08R B03R B09R B02R W04R B010R
B11R W112R B16R B15R W111R 56511
8NiceGo19N B06R W011R B02R W07R W05R W09R W01R B012R
B14R W110R B13R W18R 4646
9MCark B01R B05R B07R B08R W011R B03R W06R W04R B010R
B12R W19R B112R 3610
10Orego12 W04R B010R B01R W06R B07R W05R B011R W03R B08R W02R B09R W012R

In the table above,
   0 is a loss
   1 is a win
   J is jigo
   left superscript is the player's colour
   right superscript is the round in which the game was played
   a subscript shows how the result was determined:
      R for resignation
      T for time
      F for forfeit
      a number for the points difference after counting.
All the 0s, 1s and Js are links to the game record.

Black won 28 games and White won 32.

Zen19S vs nomiBot
Moves 191-193


In round 1, Zen19S was white against nomiBot. After move 191 in the position shown to the right, it appeared to me that the large white group in the upper right was unsettled, and White could win by keeping it alive with, say, a move at 193. But Zen19S played tenuki with 192, and nomiBot killed the group with 193. It remained dead, and Zen19S lost the game. I, and a majority of the kibitzers, thought that 192 was a blunder, uncharacteristic of Zen.
       However, we were wrong; Zen19S was already in a poor position when it played 192, which was a good move. Its real blunders had been earlier, including one which lost its group in the lower right. Zen19S's operator Hideki Kato explained "The move 192 is not a blunder, but was the best move for Zen when the winrate had already fallen to 21%. After blunder moves O3, N4 and O14, the winrate dropped from 68% to 56%, remained constant at 56%, and dropped from 54% to 16%, respectively. Clearly the first was recoverable, the second was not actually a blunder, and the third was fatal as N5 would have saved White's O5 group (M7 and P4 are then miai).

pachi vs Zen19S
Moves 72-74

In round 3, Zen19S was black against pachi. In the position to the left, move 72 threatens to kill five black stones and so save the dead white group in the upper right. However Zen19S played tenuki, allowing pachi to play 74 and win the game. 73 is clearly a blunder. Zen19S's operator Hideki Kato reported 'some floating operation error; "value" is "inf"'. He later reported "I ... checked my code and found a bug, which was added in, perhaps, last December".

Fuego19 vs CrazyStone
Moves 92-106.
98 is at 105.

In round 7, CrazyStone was black against Fuego19. With moves 92 to 104, it appeared to be trying to save its dead r6 group by chasing some stones in a ladder that did not work. But this gave it a useful ponnuki in the centre of the board; and when it abandoned the failed ladder with a tenuki at 106, it left the black group in the upper right very weak. In fact it went on to kill this group and secure the entire upper right quarter of the board as territory, and win the game. So I guess that when it played the non-working ladder, it "knew what it was doing" and was building up strength which would be effective against the upper right.

Annual points

Players receive points for the 2013 Annual KGS Bot Championship as follows:

Crazy Stone8

Details of processor numbers, power, etc.

Aya, running on 6 cores of ann i7-980X 3.3GHz 6cores +HT.
Crazy Stone, running on a 24-core PC.
Fuego, running on an 8 core Intel(R) Xeon(R) E5420 @ 2.5GHz for the first seven rounds, then on a 4 x 16-core Opteron 6376 @ 2.3 GHz.
Many Faces of Go, running on an i7-2600 (4 cores, 8 threads, 3.4 GHz.
MC_ark, running on a Core-i7 3770 3.40GHz, 4 cores (8 threads).
NiceGo19N, running on i7-2600K.
nomitan, running on a machine with 2 Xeon E5-2680 (total of 16 cores @ 2.70Ghz).
Orego, running on one node of our custom Linux cluster Fido. The node has two AMD Six Core Opteron 2427 2.2 GHz processors (12 cores total), 8 GB RAM, Centos Linux.
pachi running on 2x Opteron 6134 (15 threads) with 64GiB RAM.
Zen, running on a mini-cluster of a dual 6-core Xeon X5680@4 GHz 24GB RAM, a 6-core i7 3930K@4 GHz 16 GB RAM, a 6-core Xeon W3680@4 GHz 12 GB RAM, and a 6-core i7 980X@4 GHz 6 GB RAM computers connected via a GbE LAN. 30 cores total.