Seventieth KGS Computer Go Tournament

Sunday April 3rd, 2011

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


format30-round Swiss
board size9×9
time4 minutes plus 10/30s


The first round started at 08:00 UTC.

Result table

Zen9 pachi Erica stv valky Fuego AyaMC ManyF gomor Orego oakfo break Simpl WeakB sf9x9 Stone
B072 W113R B120R W129R W11T B114R W123R B12R W0162 B124R W14R B012R W119R B15R W115R B125R W111R B122R W128R W13R B02115 W030R B06R W118R B18R W127R W19R W11732 W11028 B12618 24535422Winner
2pachi2 W172 B013R W020R B029R
   J9 W110R B016R W125R B03R W117R B023R B02R W114R B122R B18R W119R B126R B15R    J24 B112R W027R W111R B121R W115R W11R B1624 W11816 W13018 W128T B14T 21523½314½
3EricaBot B01T W014R B023R    J9 B010R W116R B025R
B15R W112R B127R B08R W017R B16T W022R B129R B14R W118R B030R B17R W128R W12R B124R W111R B126R B113R W12112 W1334 W11562 B119R B120T 20½514302
4stv W02R B1162 W024R W13R B017R W123R W05R B012R W027R
B07R W120R B026R B19R W014R B128R B18R W021R B16R W122R B11R W129R W14R B125R W110R B130R B1192 W11344 W11150 B115T B118T 20509½301½
5valkyria B04R W112R B019R W12R B014R W022R W18R B117R W17R B020R W126R
B013R W116R B024R W110R B027R W05R B115R W19R B028R W130R B13R W129R W111R W11888 B168 W12588 B121R W123T 20492291½
6Fuego128t W05R B015R W025R W08R B019R W026R W06T B122R W029R W09R B114R W028R W113R B016R W124R
W112R B14R W117R W17R B120R B121R W123R B118R W127R B11030 W13016 W11116 B1274 W11T B13T 19505269
7AyaMC B011R W022R B028R W05R    J24 W04R B018R W130R W08R B121R B010R W127R B012R
W01R B119R W123R W115R B017R W125R W113R B114R W12R B116R B1320 W1262 B1204 B1648 B19R W129R W17T 18½483½235½
8ManyFaces1 B03R W12115 B130R W012R B127R W07R B028R W06R B022R B15R W015R W04R B017R B11R W019R B023R
B113R W016R B110R W118R W114R B120R W11132 B12574 W12646 B1974 W12438 W18R B12R W129R 18472½235½
9gomorra3 W16R B018R B011R W021R B02R W024R W01R B029R B09R W128R B030R B07R W020R B015R W117R B025R W013R B116R
W15R B13R B184 W1236 W11234 B11916 W1416 B1224 W110R B126R B114T W127T 16461½181½
10Orego12 W08R B027R B015R B011R W026R B04R W025R W03R B029R W021R B023R B013R W014R W010R B018R B05R
B17R W124R B198 W1162 B1284 W124 B1224 W13010 W114 B11740 W16R B120R B112T W119T 14438121
11oakfoam B09R B01R W013R B010R W030R B011R W018R B027R B02R W016R B014R W020R W03R W07R B024R
W1510 B022R W12924 B1838 W12111 B12834 B11226 W019T B12316 B14R W117T B125R B16T W115T B126T 13391½91
12break9 B01732 W0624 B02112 W0192 B01888 W01030 B03016 W0320 B0262 B01132 W084 B0236 W098 B0162 W0284 B0510 W122R B02924
B04T W11536 B12716 W1736 B11420 W12034 W12T B112R W124T B0118 W113T B125T 11389½79
13SimpleBot B01028 B01816 B0334 B01344 W068 B01116 W0204 W02574 B02646 B01234 W01916 B024 W0224 B03010 W0838 B02111 W02834 W14T B01536 W02716
B1544 W11634 B12944 B1710 W114T B0238 W19T B117T W124T 1037354
14WeakBot50k W02618 B03018 B01562 B01150 B02588 W0274 W0648 W0974 B02438 B0416 W0224 B014 W01740 W01226 B119T W02316 B0736 W01420 B02034 W0544 B01634 W02944
B13T W113T B118T W127T W18T B110T W121T B128T 936941
15sf9x9bot B028T W019R W015T W021R B01T W09R B029R B08R B010R W026R B06R W020R W04R B017T W025R B02T W012R B024T W0710 B014T W1238 W03T B013T W018T B027T
W15T B111T W116T B122T W030T 5363½18
16StoneGrid W04T W020T W018T B023T W03T B07T W02R B029R W014T B027T W012T B019T W06T B015T W026T W1118 B013T W025T B09T W017T B024T B08T W010T B021T W028T B05T W011T B016T W022T B130T

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.

The numbers in the table may not add up as you might expect. This may be because
some entrants played no games, and games against absent opponents treated as wins; or
because of one or more players which played one or a few games before being removed.


There is a view that for 9×9 tournaments, the komi of 7½ specified by the Chinese rules is too high, and favours White. When this was discussed earlier in the year, some potential entrants preferred a komi of 7, while others favoured remaining with 7½. I took a vote, in which those in favour of changing to 7 won by three votes to one. So for this event, we used komi of 7.

This was awkward for some MC programs, which cannot eaasily allow for three possible game results. I expected to see at least ten jigo results – if programs can count and read well, they should often find that the best result they can achieve is jigo. And if they can't count, or more likely can't read well, they should obtain a random spread of results, which on a small board should still include a significant proportion of jigos. But that was not how things worked out.

Seventeen players registered. I tried to remove IdiotBot before the start of round 1, to make the numbers even; but I was too slow, so it was included in round 1.


In round 1, SimpleBot had a bye, while valkyria had an easy win against IdiotBot.

Also in round 1, EricaBot achieved a four-point win against Zen9; but then, instead of passing, made meaningless moves, slowly, inside Zen9's territory until it lost on time. Its author told me that he had its time setting wrong, he changed this after the game.

During round 3, StoneGrid stopped making moves, and lost on time.

After three rounds, there were three players with three wins each, Zen9, valkyria, and Fuego128t. Pachi2's operator realised that it had the wrong opening book loaded, it had been failing to start in the centre when Black. He restarted it without a book, but it continued to prefer e6 to e5 for its first move as Black.

In round 4, StoneGrid joined its game but made no moves. I waited until it had lost its game on time, and kicked it off the server. I did this in the hope that the kick would wake it up and make it play moves; a kick sometimes has this effect on a bot.

After round 4, Zen9 and Fuego128t were the only players on four wins.

In round 5, StoneGrid joined its game a few seconds late (it takes bot five minutes to reconnect after being kicked). However it still did not make any moves.

Zen9 beat Fuego128t, to become the only undefeated player.

Fuego128t vs EricaBot
After move 65.

In round 6, Gomorra3 beat Zen9. This was Zen9's first loss.

Also in round 6, Fuego128t and EricaBot reached the settled position shown to the right, with all the dame filled. If they had passed at this point, it would have been jigo. But both players played on, Fuego128t rather slowly despite being short of time, and EricaBot won on time after 22 more moves.

In round 7, sf9x9bot achieved a won game against SimpleBot, but the players disagreed about the status of some groups. In the clean-up phase, sf9x9bot passed throughout, allowing SimpleBot's dead stones to be counted as alive, and lost as a result.

After round 7, EricaBot was in the lead, being the only player with six wins.

Fuego128t vs stv
Move 35.

In round 9, stv played move 35 as shown in the diagram to the left, against Fuego128t. I believe that the upper left white group is now dead, as "bent four in the corner". Both players played as if they understood this.

Also in round 9, EricaBot and pachi2 achieved a jigo.

I observed that whenever oakfoam is black, it thinks for 25 seconds before deciding to start at e5.

In round 10 EricaBot and pachi2 were paired again. This time pachi2 won.

valkyria vs Zen9
Move 50.

In its round 12 game with Zen9, valkyria played as shown to the right, building a group that is alive only in ko. It then converted this to a double ko, then killed the black group in the top right, winning the game. If this description sounds confused, I encourage you to look at the game record, which is linked to from the cross-table at the top if this page.

After round 12, valkyria was in the lead with 10 wins, pachi2 in second place with 9½, Zen9 third with 9, and EricaBot 4th with 8½.

Sometime in round 14, sf9x9bot stopped making moves. After it had lost on time in round 15 I kicked it in the hope of waking it up. This was probably effective in round 16, but it was drawn as White against StoneGrid, which had not moved since round 3, so it was impossible to tell.

After round 16, Zen9 and valkyria shared the lead with 12 points, EricaBot and pachi2 shared third place with 11½, and stv was in fifth place with 11.

In round 17 sf9x9bot played normally against oakfoam, but failed to play at all during the clean-up phase, and lost on time. I kicked the inactive StoneGrid again.

In round 18 StoneGrid returned from its kick but still did not make any moves. Sf9x9bot started its game with WeakBot50k, but stopped playing moves after move 15, and lost on time.

In round 19 The leaders, Zen9 and valkyria, each on 14 wins, played each other. Zen9 won. Oakfoam stopped playing in a won position against WeakBot50k, and lost on time.

ManyFaces1 vs Zen9
After successive passes.

In round 21 Manyfaces1 and Zen9 both passed in the position shown to the left. ManyFaces1 claimed that its group in the upper right was alive; Zen9 claimed, correctly, that it was dead. But in the clean-up phase, Zen9 passed instead of killing it, so the game was scored as a win to ManyFaces1.

After round 23, Zen9 was in the lead with 18 wins, pachi2 was in second place woth 16½ wins, and Fuego128t, stv, and valkyria were tied for third place with 16 wins each.

pachi2 vs AyaMC
After the last real move of the game.

In round 24 pachi2 and AyaMC achieved a jigo, as shown to the right. They then made hopeless invasions of each other's territory for another 60 moves, with no effect on the result. I do not know if this behaviour was connected with the jigo – I guess that if one of them had been half a point ahead, only its opponent would have tried hopeless invasions.

In round 28, sf9x9bot fell asleep again. I woke it by kicking it.

In round 29, StoneGrid'd operator connected to KGS, and realised that StoneGrid had not made a move for four hours. He woke it up, and it started playing its game against ManyFaces1, but with only 1½ minutes left. It lost.

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



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
Erica, running on an i7-950, 4 cores processor of 3.07 GHz.
Fuego, running on a Power 7 running at 3.3Ghz with 32 cores and 128 threads
Gomorra, running on a 12 core SMP machine. Cores are running at 2.67 Ghz.
Many Faces of Go, running on 4 cores each 2.3 GHz.
Oakfoam, running on an AMD Athlon(tm) 64 Processor 3000+
Orego, running on one of the five nodes of a custom Linux cluster build by PSSC Labs: the node has two AMD Six Core Dual Opteron 2427 2.2 GHz (12 cores total), 8 GB RAM, Centos Linux.
pachi, running on 64 unspecified 20-core platforms.
Stone Fighter, running on a cluster of two intel i5 750 computers and one AMD phenom II x4 965 computer, with a total of 12 cores.
StoneGrid, running on an Intel Core2 Quad Q9400 (2.66GHz)
running on Linux, 8GiB RAM, AMD Athlon(tm) 64 X2 Dual Core Processor 4000+
Steenvreter, running on an old Intel Core2, Q6600 (4 cores)
valkyria, running on a i7-860 4 Core processor at 2.80 Ghz.
running on Linux, 2GiB RAM, Intel(R) Celeron(R) M CPU 530 @ 1.73GHz
Zen, running on a Mac Pro 8 core, Xeon 2.26GHz