Google
 


  

The Shodan Go Bet

This the current plan and agreement as of Mid-December 2010. Any of this may be changed by mutual agreement. Back to the main page.

  

Competition Details

Board Size19x19 1
RulesetAGA rules (KGS)2
Komi7.53
Time Limits110 minutes each, sudden death4
VenueInternational Students House, London, U.K.
(at same time and location as the London Open Go Tournament).
229 Great Portland Street, London W1W 5PN.
Nearest station: Great Portland Street underground station.
Dates To PlayDecember 28th to 31st, 2010
Number Of GamesBest of seven 5
Computer ProgramTo Be Decided (just before the event)
HardwareThe program, at least for the first two games, will run on a notebook, i7-820QM (4 1.7Ghz cores).6

The current plan is that two games will be played each day, at 10am and 2pm, on the Tuesday, Wednesday and Thursday (December 28th to 30th), and then if still not decided, the final and deciding game will be Friday, December 31st, 2010, at 11am.

The match will be played in the Fitzroy Room; it may not be open to the public, but the games will be relayed to the Gulbenkian Room, where they can be viewed and discussed.

  • 1: Most common size played by humans.
  • 2: Originally we agreed on Tromp-Taylor rules, but there was concern people might not understand them, and that the program might not support them. AGA rules, as implemented by KGS means: area scoring, eyes in seki worth a point and situational superko (this is the only difference from BGA rules, which use natural situational superko).
  • 3:We had previously agreed on a komi of 6.5pts, but that is unusual for area scoring rules, so the more usual 7.5pt will be used.
  • 4: About an hour each seems common for serious tournaments for amateur dan players. More time would probably favour a MCTS program, but the difference would be minor. We were going to use "Fischer time control. 40 mins + 20 secs/move.", but it seems no programs currently support it. We then planned to use 90 mins each, then Canadian overtime, 20 moves in 5 minutes; but this was considered a pain to implement (especially as the computer's own clock won't be in sync). The first two games were played with 110 minutes each. For game 3 onwards the computer will get an extra 15 minutes on the clock to allow for time spent copying moves back and forth.
  • 5: Best of 10 was the original bet, but that was changed to best of 9 to avoid the possibility of a draw, and then reduced to best of 7 to avoid having to play more than two games a day.
  • 6: The hardware must be physically present, and we have agreed that a $5000 limit on the value of the hardware captures the spirit of the original bet. However there are no other limits, and it can be a cluster of computers. There were some issues arranging hardware (neither of us living in London, and a bad time of year with universities and companies mostly closed), so after the first two games we may use a remote computer that is equivalent to what could be bought within that $5000 budget.

© Copyright 2009, 2010 Darren Cook <darren@dcook.org>

Revision History
2009-05-26: First public release
2010-06-20: Updated with venue information.
2010-06-27: Updated ruleset description
2010-12-24: Update with correct times, time limits and latest hardware description.