Sixteenth KGS Computer Go Tournament

Sunday July 1st, 2006

These results also appear on official KGS pages: Formal Division, Open Division which link to the game records.

Rules

Formal division:
9x9 board
Chinese rules, komi 7½
13 minutes each absolute time
Open division:
13x13 board
Chinese rules, komi 7½
18 minutes each absolute time

Format

Formal, eight-round Swiss; Open, six-round Swiss.

Times

The first round started at 17:00 UTC for the Formal and 17:05 for the Open division, subsequent rounds started at thirty-minute and at forty-minute intervals for the two divisions.

Results

As usual, the tournament was held in two divisions, Formal and Open, with more restrictive entry conditions for the Formal division.

Formal Division   9x9

placenamewinsSOSSoDOS
1stCrazyStone82323
2ndAyaBot4336
3rdMoGoBot3356
4thantbot9x91373

Open Division   13x13

placenamewinsSOSSoDOS
1stStoneCrazy61717
2ndSimpleBot51913
3rdHouseBot4165
4thWeakBot50k4165
5thFomalhaut2201
6thIdiotBot2201
7thHBotSVN1170

The "real" names of the bots listed above, and of their programmers, are listed here: programs which have registered for KGS Computer Go Tournaments.

Formal division

We welcomed a new player to the Formal division: MoGo. It is the work of a group of French programmers: Rémi Munos, Pierre-Arnaud Coquelin, Yizao Wang, Olivier Teytaud, and Sylvain Gelly.

In its first game, against antbot9x9, it was white and played the sequence shown. I was impressed by this, it appears to be able to read and to know what it is doing. However, it subsequently played some strange moves, and antbot9x9 eventually won by 4½ points. Meanwhile CrazyStone, following its recent success at the 9x9 Go in the 11th Computer Olympiad in Turin, beat AyaBot by half a point.
 

In round 2, CrazyStone and MoGoBot gave me as organiser some work when they both passed in this position, and then argued about the status of the large white group. This implies three separate errors - Black passed when there is a move to be made at the lower right; so did White; and one of them (I can't tell which) claimed that White's seven-eyed group was dead. So I had to count the result myself. At first I got my arithmetic wrong, but the players persuaded me to reconsider, and I realised that White has 37 points and Black has 39, so with komi White wins.

After eight rounds, CrazyStone was undefeated, though most of its wins were close. Its winning margins against AyaBot were ½, 3½ and ½; and against antbot9x9 2½ and 3½.

Open division

HBotSVN was readmitted to the event. It had been barred from the June tournament after what I considered unreasonable behaviour in the May tournament, disputing the status of its opponents' groups when they were clearly alive. Its author assured me that a problem causing this had been fixed, and I was able to verify this.

Another new player, Razorbot2, was registered for the Open division, but was unable to play. It withdrew from the event after two rounds.

HBotSVN lost its first game on time, in a won position, because its time setting was wrong. Its programmer corrected this before the second round, and it did not lose any more games on time.

There is a question about what changes to a program (and its settings) should be allowed during the course of a tournament. My view is that that changes prompted by things that happen during the tournament (e.g. how a game is going, or which opponent may be next) are not allowed. However changes which would be desirable regardless of how things are going (like correction of wrong time settings) are allowed,


Forfeited games

The following "forfeited game" records are available by clicking on them:
CrazyStone-MoGoBot.sgf
MoGoBot-antbot9x9.sgf

Other game records are available from the KGS pages: Formal Division. Open Division.