| Formal division | Open division | |
|---|---|---|
| board size | 9x9 | 13x13 |
| rules | Chinese | Chinese |
| komi | 7½ | 7½ |
| time | 28 minutes absolute | 43 minutes absolute |
Formal division: six-round Swiss.
Open division: four-round Swiss.
The first round started at 17:00 UCT for the Formal and 17:10 for the Open division.
As usual, the tournament was held in two divisions, Formal and Open, with more restrictive entry conditions for the Formal division.
Formal Division 9x9
|
Open Division 13x13
|
The "real" names of the bots listed above, and of their programmers, are listed here: programs which have registered for KGS Computer Go Tournaments.
I was pleased to welcome a version of David Fotland's program Many faces of Go, the first time it has played in one of these events. ManyFaces1 is not the currently available version 11 of Many faces; it is a new experimental version, doing alpha-beta full board search, 5 plies deep. He warned me that it might be buggy, it might not support kgsGtp correctly, and he would be unable to babysit it after the first round. He expects that a debugged version of it will play in future KGS events.
GNU played in both divisions. The version in the Open division was the standard development version. The one in the Formal division, MonteGNU, was a version modified to use a fairly standard and unsophisticated UCT search restricted to choosing among the moves normal GNU Go move generation would consider.
In round 1, MonteGNU disappeared from its game with Orego with connection problems. Five minutes later Orego was removed from the game by the "five-minute rule". MonteGNU, and its operator Gunnar Farnebäck, managed to reconnect, and Orego's operator Peter Drake was able to get it back into the game too; so the game then continued normally.
In their round 2 game, MonteGNU and indigoBot both showed an inability to understand bent-four-in-the-corner.
In round 1, GNU had the same connection problem as MonteGNU did in the Formal division, resolved in the same way.
ManyFaces1 obtained a won game against IdiotBot, but after both had passed, it appeared unable to do the GTP equivalent of clicking the "Done" button. Eventually I assigned the win to ManyFaces1. This game is therefore shown on the KGS page as forfeited by IdiotBot; but this is through no fault of IdiotBot's.
In round 2, ManyFaces1 did the same again, winning against AyaBot2 and then failing to click "Done". Again I assigned the win to ManyFaces1.
In round 3, when the game between HBotSVN and HouseBot was ending
in the position shown, there was some concern that neither of the players
knew how to score a seki, and my intervention might be needed. We need not
have feared. With the next two moves, Black, and then White, filled their
own eyes in the seki. They then recovered their sanity; Black captured the
five white stones, and White played on the vital point of the nakade.
ManyFaces1 connected to its game with MoGoBot13, but failed to make a move. In the hope of waking it up, I kicked it. It reconnected and rejoined the game, but still it did not move. Eventually it lost on time.
In its round 4 game against Mango, ManyFaces agin timed out without making a move.
HouseBot obtained a won position against IdiotBot. However it does not implement the kgsGtp clean-up instruction, so IdiotBot's dead stones were treated as alive by the server, which counted the result as a win for IdiotBot.