pass and unsettled groups remain on the board, they are assumed to alive.
The tournament director will resolve any disagreement between programs on
the status of groups after the game is over.
Participants will make a game record.
Each program will have one hour to make 125 moves. Games will be played
until both programs pass or one program has used up its
hour of playing time. If less than 125 moves have been played by the
program
that ran out of time it loses. Otherwise, if there
is time left in the round the game will continue.
Any time after move 250 the referee may
determine the outcome of the game in order to allow the next round to start.
If a program crashes it can be restarted, but its clock continues to run.
Any number of program crashes is allowed as long as the time limit is
not exceeded. If the crash is due to something unrelated to the go
playing algorithm, such as low memory, out of stack space, or similar, it
may be recompiled or relinked, or the computer may be rebooted with
different memory parameters while the clock is running. Go playing
code or pattern
database changes are not allowed during a round.
No changes to the program or its parameters may take place during a round,
but changes are allowed between rounds. The program may request the time
left on its clock periodically and adjust itself accordingly. The operator
may not adjust parameters to make the program faster if it is running out of
time, except to enter a low on time command or make a single
adjustment of a time left, or playing strength parameter
when there is 10 minutes left on
the clock.
If a game is interrupted due to power failure or hardware failures,
and the programs can't be reset to the position when the failure happened,
if more
than 150 moves have been played (75 each) and the outcome can be determined,
the referee will adjudicate the result. Otherwise the game will be played
over.
If there is enough time and a small enough number of entries then the
tournament will be round robin with each program playing each other one
once. We should have time for 6 rounds, but the actual number of rounds and
schedule will be determined when the contest begins.
Otherwise it will be a Swiss style tournament.
First round pairings will be determined
by past results of the programs or by chance. In each round programs with
the same number of wins will be paired whenever possible. No program will
get more than one bye. A bye counts as a win. The same opponents will not
be paired twice. The tournament referee has final say on the pairings.
Tournament results are determined first by number of wins, then by sum of
defeated opponent's scores, then by sum of opponent's scores, then by
head to head competition.
Programs are encouraged to implement the Standard Computer Go Modem Protocol
to allow them to play one another without human intervention. If a program
does not implement the protocol, moves will be typed by hand between it
to its opponent. The protocol spec and sample code from Many Faces of
Go and Nemesis are available from David Fotland or Bruce Wilcox. The
spec has been published in 'Computer Go'. It is also available from the
IGS archives at igs.nuri.net.
Using the modem protocol gives each program more time, and speeds the
rounds since operator time is removed.
If the Modem protocol is available,
by mutual agreement and with consent of the tournament director, the
game can be played without a clock and without updating the position on
a separate go board. If such a game is interrupted by a crash, power
failure,
or failure of communication between programs, and the tournament director
cannot reconstruct the correct position, then the game will be played over
using clocks and a go board.
-Original Message-
From: [EMAIL PROTECTED] [mailto:computer-go-
[EMAIL PROTECTED] On Behalf Of Peter Drake
Sent: Wednesday, December 05, 2007 8:17 AM
To: computer-go
Subject: Re: [computer-go] programs at the US Go Congress
I'm in charge of organizing a computer Go event at this year's
Congress. Right now I'm trying to get one or more local companies to
donate machines and/or prize money.
I really doubt we can subsidize travel.
Stay tuned...
Peter Drake
http://www.lclark.edu/~drake/
On Dec 4, 2007, at 6:23 PM, David Doshay wrote:
Hi,
What should we be doing to get programmers to bring their bots to
the Congress in Portland in 2008?
The AGA is formally not that program friendly, but there can be
events for bots, and hopefully events for humans against bots. I am
sure that there will be tournaments that will not include programs,
but if we start now I am sure that we can make it interesting for
all. Chuck Robins, one of the TDs, said that he is willing to have
tournaments that allow computers.
The recent thread on testing the strong programs and what ratings
they might have has me thinking that it might be good to have
various