Eightieth KGS Computer Go Tournament

Sunday February 5th, 2012

These results also appear on an official KGS page.

Rules

format24-round Swiss
board size9×9
rulesChinese
komi7
time4 minutes plus 10/30s

Times

The first round started at 16:00 UTC.

Result table


PlaceNamecross-tableWinsSOSSoDOSNotes
stv Zen9 pachi ManyF AyaMC MyGoF oakfo Orego
1stv
X
W04R B173 WJ15 B017R W019R B15R W111R B116R W123R B12R W110R B118R W124R W13R B18R W112R B121R W19R B11313 W12023 B11R W114R B16R W122R 20½296½233Winner
2Zen9 B14R W073 BJ15 W117R B119R
X
W16R BJ9 W118R BJ21 W0222 W034 B0819 W120R W15R B010R W116R B123R W12R BJ24 W112R B0135 W111R B114R 16310169
3pachi2 W05R B011R W016R B023R B06R WJ9 B018R WJ21 B1222
X
W01R B112R W017R W02R B113R W115R B124R B178 W11422 W18R B120R W13R B110R W119R 14302½117½
4ManyFaces1 W02R B010R W018R B024R B134 W1819 B020R B11R W012R B117R
X
W04R B09R W014R B022R B1522 W115R B1194 B16R WJ11 B116R W123R W17R B113R W121R 13½296½126
5AyaMC B03R W08R B012R W021R B05R W110R B016R W023R B12R W013R B015R W024R B14R W19R B114R W122R
X
W1610 B111R W018R W17R B119R B11R W117R B120R 13315132
6MyGoFriend B09R W01313 B02023 B02R WJ24 W078 B01422 W0522 B015R W0194 B0610 W011R B118R
X
WJ3 B14R WJ10 B117R WJ21 B122R B18R W112R B116R W123R 1125750
7oakfoam W01R B014R B012R W1135 B08R W020R W06R BJ11 W016R B023R B07R W019R BJ3 W04R BJ10 W017R BJ21 W022R
X
B15R W09R B115R W018R B124R 726728
8Orego12 W06R B022R B011R W014R B03R W010R B019R B07R W013R B021R W01R B017R W020R W08R B012R W016R B023R W05R B19R W015R B118R W024R
X
4274½15

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.

The numbers in the table do not add up as you might expect. This is because they include points for wins against the absent MCark, and for byes while it was included in the tournament.

Results

Near the end of round 1, MCark stopped moving in its game with Zen9, and lost on time.

Throughout rounds 2, 3, and 4, MCark was not connected to the server.

In round 3, oakfoam, as black, achieved a jigo against MyGoFriend.

In round 4, after MCark has lost its fourth game on time, I gave up hoping that its operator would reconnect it, and removed it from the tournament, leaving an even number of players.

In round 8, pachi2 was two minutes late for its game with oakfoam. It still won.

In round 10, oakfoam, as black, achieved another jigo against MyGoFriend.

In round 11, oakfoam, as black, achieved a jigo against ManyFaces1.

AyaMC vs stv
Moves to end.

In round 12, AyaMC, as Black, appeared to miss two snapbacks playing against stv, as shown to the right. (The first snapback: if Black captures at f5 for move 51, its group will be captured. The second snapback: if instead Black captures at d3 for move 51, White cannot effectively give atari at f5.)
    But in fact it may not have missed the snapbacks; rather, its move 47 is irrelevant as its position was already hopeless. If instead it had blocked at c6, White would still play at 48, capturing and winning; while if it had connected at e5, White would push through at c6, securing more than half the board.

oakfoam vs Zen9
At game end.

In round 13, Zen9 (as Black) and then oakfoam passed in the position shown to the left. Zen has won: the 14-stone white group in the bottom left is dead (Black can play at a2 and then b1, almost filling White's eyespace with a bulky-five shape). However they disagreed about the status of the 14-stone group, and the game entered the clean-up phase. Oakfoam again passed, and the game was immediately declared over, with all the stones on the board counted as alive. The game was therefore counted as a win for oakfoam.
    It seems to me that Black was never given a chance to show that it could capture the dead group. I shall check, and if my understanding is correct, I shall report this to KGS programmer 'wms' as a defect in the clean-up mechanism.

The round 20 game between stv and MyGoFriend was also mis-scored. Two dead white stones were accepted by both players as alive. This increased stv's margin of victory, but did not affect the result.

In round 21, oakfoam, as black, achieved a third jigo against MyGoFriend. I suspect that MyGoFriend was trying to win by half a point, not understanding that the komi was on 7, not 7½.

MyGoFriend vs Zen9
Moves 49-51.

In round 24, the last three significant moves are as shown to the right. My understanding of them is as follows.
    Move 49: Zen9 as Black had no way to avoid losing, so this move was as good as any.
    Move 50: This move is worthless. But MyGoFriend thinks the komi is 7½, and therefore believes it has won whatever it does.
    Move 51: This secures a jigo for Zen9.

Conclusions

If you believe, as most MC-based programs do, that a win by half a point is as good as a win by thirty points, then it is advisable to understand what the komi is.

Annual points

Players receive points for the 2012 Annual KGS Bot Championship as follows:

Steenvreter8
Zen5
pachi3
Many Faces of Go2
Aya1


Details of processor numbers, power, etc.

AyaMC
Aya, running on 6 cores of an i980X, at 3.3GHz.
ManyFaces1
Many Faces of Go, running 4 cores (8 threads) of an i7-2600.
MCark
MC_ark, running on a Core-i7 2600K 3.40GHz*4core (8 threads)
MyGoFriend
MyGoFriend, running on a 20-core system: two i7s each with 6 cores and two i7s each with 4 cores, all about 3.3 mhz.
oakfoam
Oakfoam, probably running on an Intel Xeon E5645 @ 2.40 GHz
Orego12
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.
pachi2
pachi, running on 64 platforms, each x86 64 bits, 32 GB ram, using 22 cores of each (total 1408 cores), giving about 1500 playouts/s/core at the beginning of a 19x19 game.
stv
Steenvreter, running 46 threads each at 2.2 GHz, on a system whose use was generously provided by the Maastricht games and AI group.
Zen19
Zen, running on a Mac Pro 8 core, Xeon 2.26GHz.