| Formal division | Open division | |
|---|---|---|
| format | 5-round Swiss | 5-round Swiss |
| board size | 19×19 | 19×19 |
| rules | Chinese | Chinese |
| komi | 7½ | 7½ |
| time | 28 minutes absolute | 28 minutes absolute |
The first round started at 13:00 UCT for the Formal and 13:10 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 19×19
|
Open Division 19×19
|
The "real" names of the bots listed above, and of their programmers, are listed here: programs which have registered for KGS Computer Go Tournaments.
This is the first of these events in which 'break' written by YangYun, actually played. It played in the Formal division, using the name 'break19'.
CrazyStone registered to play in both divisions; then withdrew again when Rémi realised he would be away on the day of the event.
IdiotBot was removed from the draw to make the number in the Open division even.
Round 1 was unremarkable. MoGoBot1 beat FirstGoBot (which has improved greatly in recent months). GNU killed all break19's stones.
In round 2 break19 timed out in a lost position against FirstGoBot. MoGoBot1 achieved a characteristic UCT-style half-point win against GNU, with all the dame filled and six minutes left on its clock. It then went to sleep and timed out.
In round 3 FirstGoBot resigned to GNU after losing a group. At move 361 of the MoGoBot1/break19 game, all the dame were filled and MoGoBot19 was well ahead, with over two minutes on its clock while break19 had three seconds. But break19 wisely passed, and MoGoBot19 timed out.
In round 4 FirstGoBot was again drawn against GNU. It had an early success, killing a 19-stone central group, and may have been winning at that point; but it subsequently made worse mistakes than GNU and lost.
MoGoBot1 killed all of break19's stones (well, break19 helped by filling one of the two eyes in its last remaining live group). Then both players passed. However they disagreed about the scoring – break19 claimed that all its dead stones were alive. MoGoBot19 had only three seconds left on its clock, and in the confirmation phase, while break19 passed, MoGoBot19 concentrated on filling its own territory instead of getting the dead white groups off the board, and lost on time while doing so.
In round 5 GNU easily killed all of break19's stones, and removed them from the board with seven minutes to spare. FirstGoBot resigned in a lost position against MoGoBot1.
In round 1 MoGoBot2 and hb04 achieved a position with all the dame filled, and both passed. However the programs now refused to proceed with the game, neither agreeing nor disagreeing about which stones were alive. As MoGoBot2 clearly had a won position, I assigned the win to it; and then I killed the game, so as to free up the players to play in the next round.
In round 2 the final stages of the SimpleBot-hb04 game were odd. When all the dame had been filled, SimpleBot was over 80 points ahead. SimpleBot then passed repeatedly while hb04 filled its own eyes. Then hb04 started playing inside SimpleBot's territory: this woke up SimpleBot, and it started capturing the dead stones, including hb04's previously live groups, once hb04 had filled all but one of their eyes. With move 584, SimpleBot captured hb04's last group. Hb04 played inside the space thus created, and after responding once, SimpleBot went to sleep, and lost on time.
MoGoBot2 achieved a won position against WeakBot50k, with all the dame filled. Then, after some pointless moves, MoGoBot2 fell asleep and lost on time.
WeakBot50k had won its first-round game against HBotSVN, and was now leading the tournament; the only other program with two wins was MonteGNU, and WeakBot50k was ahead of it on tie-break. Leading a tournament after two rounds must be a record for WeakBot50k.
In round 3 MoGoBot2 achieved a won position against MonteGNU, with all dame filled; and then once more lost on time. There was speculation that the bug causing this in in its garbage collection; the bug has has not been noticed before, as it normally plays with faster time limits, and so does not have time to make so much garbage. I learned later that the bug is in fact in its scoring code.
MoGoBot2 won its games in rounds 4 and 5 without any time problems.
In round 5 HBotSVN crashed before its second move in its game against hb04, and so lost on time.