Fifty-fifth KGS Computer Go Tournament

Sunday January 10th 2010

These results also appear on an official KGS page which links to the records of all the games.

Rules

format18-round Swiss
board size9×9
rulesChinese
komi
time4 minutes plus 25/20s

Times

The first round started at 16:00 UTC.

Result table


placenamewinsSOSSoDOS
1stFuego16187157
2ndZen915178139
3rdpachi12181100
4thAyaMC1118087
5thCzechBot1117881
6thgomorra9916551
7thManyFaces1915948
8thkiseki915349
9thPNUGo712312
10thSimpleBot616227
11thbreak941370
12thWeakBot50k01480

The numbers in these tables do not add up as you
might expect. This is because a player that does
not show up for a round still receives half a point
for the "bye", while its opponent received a full point.

PNUGo is a build of GNU Go, entered by Petr Baudiš.

CzechBot is a build of MoGo, also entered by Petr Baudiš.

Aloril entered his usual collection of weaker bots, and allowed me to remove IdiotBot if it would make the numbers even.

Results

All 13 registered players were present at the start of the tournament. However I chose not to remove IdiotBot to make the numbers even and avoid byes, until I had evidence that they were all capable of joining their games and making moves in them.

kiseki vs SimpleBot
At the end of the game.

So in round 1, there was a bye, which happened to be assigned to break9. All the other players showed that they were able to play; so I removed IdiotBot from the draw for subsequent rounds.
       When the game between kiseki and SimpleBot SGF stopped with two passes, in the position shown to the right, all the dame had been filled, the status of all the groups was clear, and kiseki had a won position. However kiseki does not know how to handle the clean-up phase (it had therefore captured all Black's dead groups during the game). Therefore the server treated it as claiming the dead white stones as alive, and the game was resumed. If kiseki had passed throughout the resumption phase, it would certainly have won; but whenever it had some stones captured, it played back into the space, more slowly than it could afford, and lost on time.

AyaMC vs pachi
Black's move 43.

The round 7 game between AyaMC and pachi SGF was a good one. It ended in the position shown to the left, when AyaMC accepted that all its upper-edge stones were dead, and resigned.


Zen9 vs Fuego
Moves 32 and 33

Zen9 apeared (to me) to be winning its round 9 game with Fuego SGF, when it made move 32 as shown in the diagram to the right. This achieves nothing (unless Black answers it wrong). When Black answered sensibly at 33, Zen9 resigned. Black has a push at f7 followed by a cut at g8, which will secure a win.


SimpleBot vs break9
At the end of the game.

Also in round 9, SimpleBot and break9 SGF both passed in the position shown to the left, agreed that all the stones on the board were alive, and allowed it to be counted as it is, giving the win to White by 18½ points. In fact the white group in the upper left is dead, and if the players had recognised this it would have been counted as a win to break9 by 3½ points.


gomorra9 vs pachi
At the end of the game.

In round 12, gomorra9 and pachi SGF both passed in the position shown to the right. White's upper left group is dead, and it believed, with good reason, that it had lost the game. When gomorra believes that the game is hopeless, it passes, three times. It is less clear to me why pachi passed. Pachi accidentally had its clean-up mode switched off, and gomorra9 continued to pass in the clean-up phase, so the position was counted as is, with all stones alive. Black is one point ahead on the board (the "territory" is indicated by letters b and w), so with komi of 7½, gomorra9 won by 6½.


gomorra9 vs pachi
At the end of the game.

In round 15, gomorra9 and pachi SGF were again playing, with the colours unchanged. Again, they both passed well before the game was over, in the position shown to the left. This time, they disagreed about the status of at least one group, so there was a resumption, in which they both passed again. The game was counted as a win to Black; I don't know how it was counted this way, but Black did deserve to win the game.


break9 vs Fuego
At the end of the game.

Also in round 15, break9 and Fuego SGF both passed in the position shown to the right. They disagreed about the status of at least one group, so there was a resumption, in which Black played at a, White played at b, and then both passed leaving dead groups of both colours on the board. The game was now scored as a win for Black.


The Clean-Up Phase

I note that five of the seven diagrams above show where something went wrong at the end of the game, causing it to be scored wrongly, or risking causing it to be scored wrongly. I encourage people entering these events to submit programs which support the game-end clean-up correctly.

 

"the Empty Triangle"

The tournament was watched by "chid0ri", the creator of the Empty Triangle series of on-line Go cartoons. She drew a cartoon about computer Go, no. 54 in the series, and currently the one at the Empty Triangle.


Processor numbers, power, etc.

AyaMC
Aya, running on Xeon 2.6GHz 8cores
break9
break, probably running on a single processor Intel(R) Celeron(R), 1.7Ghz
CzechBot
MoGo, unstated but probably running on double-core AMD Athlon(tm) 64 X2 Dual Core Processor 4800+ (2.5GHz).
Fuego
Fuego, running on an 8-core Intel Xeon 2.5Ghz, 8GB of memory.
gomorra9
gomorra, running 8 Threads on an Intel Core i7 Lynnfield (2.8 Ghz)
IdiotBot
running on Linux, 4GB RAM, AMD Athlon(tm) 64 X2 Dual Core Processor 4000+
kiseki
running on a Windows PC with a core2quad 2.4GHz.
ManyFaces2
Many Faces of Go, running on a quad core Q8200, 2.3 GHz.
pachi
pachi, unstated but possibly running on four cores of i7 920 @ 2.67GHz.
PNUGo
GNU Go, platform not specified but not large
SimpleBot
running on one processor of a 4GiB RAM, AMD Athlon(tm) 64 X2 Dual Core Processor 4000+, shared with WeakBot50k
WeakBot50k
running on one processor of a 4GiB RAM, AMD Athlon(tm) 64 X2 Dual Core Processor 4000+, shared with SimpleBot
Zen9
Zen, running on a Mac Pro 8 core, Xeon 2.26GHz.