Seventy-second KGS Computer Go Tournament

Sunday June 5th, 2011

These results also appear on an official KGS page.


format8-round Swiss
board size19×19
time29 minutes plus 10/30s


The first round started at 08:00 UTC.

Result table

Zen19 ManyF pachi Czech AyaMC Stone PNUGo Orego GoKno Simpl break WeakB
W12R W11R B18R B13R W15R B16R W14R B17R 83737Winner
2ManyFaces1 B02R
W05R B14R B18R W13R B1733½ B11R W16R 63723
3pachi2 B01R W08R B15R
B17R B16R W14R W12124½ W13142½ 63721
4CzechBot W03R W04R W07R
B11R W12T B15R W16128½ B08 43615
5AyaMC B05R W08R W06R
B01R B12R B13287½ B14190½ B1777½ 4339
6StoneGrid W06R B03R W01R
W15 W14R W08T B17193½ W12302½ 43110
7PNUGo B04R W0733½ B02T W11R B05
B13R B18353½ W16328½ 4319
8Orego12 W01R W05R W02R B04R
W18R W1728½ B1639½ B1363½ 4268
9GoKnot W07R B06R B04R W03R B08R
B11 B12159½ W1518½ 3335
10SimpleBot B02124½ B06128½ W03287½ B18T B0728½ W01
B15T W14226½ 3276
11break19 B03142½ W04190½ W07193½ W08353½ W0639½ W02159½ W05T
B11 1291
12WeakBot50k W18 W0777½ B02302½ B06328½ W0363½ B0518½ B04226½ W01

In the table above,
   0 indicates a loss
   1 indicates a win
   J indicates jigo
   a superscript indicates the round in which a 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.

Twelve players registered. We welcomed one new player, GoKnot-isGo, by Jacques Basaldúa. GoKnot-isGo, whose KGS username is GoKnot, is derived from his older engine, QYZ. It is based on MC with MoGo-style patterns, RAVE, progressive widening, and a new version of M-eval.


Zen19 vs pachi2
Moves 169-173

In round 1, Zen19 and pachi2 payed the sequence shown to the right. If I saw two humans of my own stregth play like this, I would assume that White had "missed" the snap-back set up by move 171. But move 170 is large: it is possible, for all I can tell, that the whole sequence was sensible.

Zen19 (White) eventually won this game.

Orego12 vs CzechBot
Moves 46, 47

In round 5 CzechBot, as Black, extended as shown to the left, and was blocked by Orego12. My impression is that the block is obvious, and that the exchange of 47 for 48 is clearly bad for Black. Quite weak human players use a tree-pruning heuristic "if an exchange leaves me in a worse position than before, then don't play it".

Also in round 5, GoKnot achieved a won position against WeakBot50k, and after two passes the players failed to agree on the status of some stones. In the clean-up phase, GoKnot did nothing for two minutes, then woke up and played fast to win the game.

Meannwhile StoneGrid and PNUGo both got into overtime, and played slowly enough to delay the start of round six by three minutes. Fortunately, the KGS tournament scheduler handles this correctly, starting a new round as soon as the scheduled time has been reached and all the games from the previous round are over.

In round 8, CzechBot got a won position against WeakBot50k, but then, with 3 seconds of main time left on its clock, started passing, allowing its opponent to capture some of its groups, and lost the game. It acted as if it was unaware of the overtime. It did exactly the same against WeakBot50k in the May KGS bot tournament.

Also in round 8 pachi2 was again drawn against Zen19, and again lost. This was the only repeat pairing of the tournament. It was unfortunate for pachi2: Zen beat pachi2 (twice) and ManyFaces1, and pachi2 beats ManyFaces1, but ManyFaces1 came above pachi2 in the final list, thanks to the SoDOS tiebreak. (The SoDOS difference was a consequence of pachi2's first-round loss to Zen, which caused it to be drawn against weak opponents in rounds 2 and 3.)

Players receive points for the 2011 Annual KGS Bot Championship as follows:

Many Faces of Go4


Details of processor numbers, power, etc.

Aya, running on 6 cores of an i980X 3.3GHz
break, probably running on a single processor Intel(R) Celeron(R), 1.7Ghz
a development version of MoGo that has been current as of 2010-10-06, running on double-core AMD Athlon(tm) 64 X2 Dual Core Processor 4800+ (2.5GHz).
GoKnot-isGo, running on 4 cores of an i7 at 3.2 GHz
Many Faces of Go, running on a 12 core Xeon; using pondering.
Orego, probably running on one of the five nodes of a custom Linux cluster built by PSSC Labs: the node has two AMD Six Core Dual Opteron 2427 2.2 GHz (12 cores total), 8 GB RAM, Centos Linux.
running on one processor of a 4GiB RAM, AMD Athlon(tm) 64 X2 Dual Core Processor 4000+, shared with WeakBot50k
StoneGrid, running on an Intel Core2 Quad Q9400 2.66GHz
running on Linux, 2GiB RAM, Intel(R) Celeron(R) M CPU 530 @ 1.73GHz
pachi, running on 32 unspecified 22-core [?] platforms.
GNU Go, unspecified platform
Zen, running on a Mac Pro 8 core, Xeon 2.26GHz