Sixty-second KGS Computer Go Tournament

Sunday August 15th 2010

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

Rules

format	15-round Swiss
board size	13×13
rules	Chinese
komi	7½
time	9 minutes plus 25/30s

Times

The first round started at 08:00 UTC.

Result table

place	name	wins	SOS	SoDOS
1st	Zen13	12	134	102
2nd	ManyFaces1	11	140	94
3rd	EricaBot	11	125	86
4th	pachi	10	126	72
5th	CzechBot	10	126	69
6th	PueGo	10	111	57
7th	Orego	7	110	34
8th	SimpleBot	7	97	35
9th	PNUGo	6	112	25
10th	kiseki	6	110	28
11th	kiaibot	6	105	21
12th	WeakBot50k	4	101	14
13th	IdiotBot	3	90	6
14th	break13	2	88	11

Fourteen players registered, and all fourteen played in every round.

We welcomed a new player, kiaibot, by Matthew Woodcraft. It uses MCTS with very light playouts.

Apology

I apologise to all the entrants. I announced the time limits as "Canadian Overtime of 10 moves in 30 seconds", I stated them like that at http://www.weddslist.com/kgs/future.html, and I intended to set them like that. However I in fact set them to Canadian overtime of 25 moves in 30 seconds, a setting which I know causes players in East Asia to lose on time because of lag, however fast they play their moves. I only realised my mistake when it was pointed out after the tournament had started.

Some players, including kiseki, suffered losses on time through my mistake.

Results


break13 vs CzechBot
After consecutive passes.

In round 1, the game between break13 and CzechBot SGF ended in the position shown to the right. The players then disagreed on the status of the marked stones, and the game entered the clean-up phase. Both players passed again. I assume that CzechBot knew it could kill the three stones, but felt no need to do so as it was ahead even with the marked stones counted as alive.

In round 2, EricaBot obtained a won game against SimpleBot, and began rapidly filling its own teritory and removing its opponent's dead stones. But, with 19 seconds left for six moves (which should not have been a problem for it) it stopped responding, and lost on time.


Zen13 vs ManyFaces1
After consecutive passes.

In the round 3 game between Zen13 and Manyfaces1 SGF, ManyFaces1 resigned in the position shown to the left. This resignation shows that it "understands" that its group in the top right corner is dead.

In round 5, pachi beat Zen13 by killing a group which (I think) should have lived in the top right, and saving a group which (I think) should have died in the bottom left SGF.

In round 6 WeakBot50k and kiaibot handled a clean-up phase correctly. Throughout the tournament, I was impressed by the behaviour of kiaibot. It consistently made legal moves well within the time limit; it resigned whenever it had clearly lost; it consistently beat IdiotBot, WeakBot50k, break13, and even SimpleBot; and it always handled the game end, including any clean-up, rapidly and correctly. Admittedly its moves were not great; but as director of these events, I am more impressed by good behaviour than by good moves.


EricaBot vs Zen13
After the end of the clean-up phase.

Also in round 6, the game between EricaBot and Zen13 ended with white (EricaBot) half a point ahead SGF. It then went into the clean-up phase, in which both players played on, removing dead groups, util they both passed in the position shown to the right. White (with 14 moves to make in 24 seconds) has failed to remove a dead stone, but black (with 36 seconds of main time left) has failed to remove the "moonshine life" group in the lower right. This did not affect the result, but it was strange.

In round 7, kiseki timed out in a won position against WeakBot50k, during the clean-up phase and while in overtime.


Orego vs PNUGo
Move 39.

In round 11, PNUGo played the move shown to the left in its game with Orego SGF. This established a seki, which both players understood, so it was still seki at the end of the game.

In round 13, break13 made one move normally, and played no more moves until it had only one second of main time left. Then it started to play again, but lost on time. It showed similar but less extreme misuse of time in the next two rounds.

At the start of round 15, Zen13 and ManyFaces1 were tied for first place, each on 11 wins. They were drawn against each other, in a game which would decide the winner of the tournament SGF. Zen 13 won.

Also in round 15, kiseki again lost on time in a won position, while in overtime during the clean-up phase.

Details of processor numbers, power, etc.

break13: break, probably running on a single processor Intel(R) Celeron(R), 1.7Ghz
CzechBot: MoGo, running on double-core AMD Athlon(tm) 64 X2 Dual Core Processor 4800+ (2.5GHz).
EricaBot: Erica (with no opening database), running 8 cores, each of 2.27 GHz, Windows 7.0
IdiotBot: running on Linux, 2GiB RAM, Intel(R) Celeron(R) M CPU 530 @ 1.73GHz
kiaibot: Kiai, running on four cores of a Core i7 860 (2.80GHz).
kiseki: kiseki, running on a Windows PC with a core2quad 2.4GHz.
ManyFaces1: Many Faces of Go, running on four cores of a four Core2 Quad 2.3 GHz.
Orego: Orego, unspecified but probably running on 2 x 3 GHz Dual-Core Intel Xeon
pachi: pachi, running on a 8-core i7 920
PueGo: Fuego, running on another 8-core i7 920
PNUGo: GNU Go, unspecified platform
SimpleBot: running on one processor of a 4GiB RAM, AMD Athlon(tm) 64 X2 Dual Core Processor 4000+, shared with WeakBot50k
WeakBot50k: running on Linux, 2GiB RAM, Intel(R) Celeron(R) M CPU 530 @ 1.73GHz
Zen13: Zen, running on a Mac Pro 8 core, Xeon 2.26GHz.