[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [computer-go] 2nd KGS Computer Go Tournament



In message <200505121616.j4CGGn9F007999@xxxxxxxxxxxxxxxxx>, Gunnar Farnebäck <gunnar@xxxxxxxxxxxxxxxxx> writes
Nick wrote:
I have only once had to deal with a situation where an unsettled group
was left on the board at the end of a computer/computer game.  This was
the game Aya (the same Aya that won the recent KGS event) versus Indigo,
in the MSO, London, 2000.  The unsettled 78-stone group belonged to
Indigo, and Indigo's owner conceded (as was evident from the play) that
Indigo thought it was dead.
and, in another message:
Or, we may be referring to "real" disputes, where the true status of a
group is difficult (for a 20-kyu) to assess, and after both bots have
passed, there is a genuine disagreement about the status of a group.
Now I have never heard of this happening.  Bots tend to defend their
groups that might be weak, and to attack their opponent's groups that
might be weak, and one way or another the situation gets resolved.  Even
with Japanese rules in force I have never heard of a "real" dispute
happening;  with the Chinese rules that we use for KGS bot tournaments,
it is far less likely.
The GNU Go versus Indigo game from the 2004 Computer Olympiad might
qualify as a real dispute. The white group in the lower left corner
changed status from dead to unsettled when GNU Go filled a liberty
with the very last move. GNU Go thought it was still dead and would
have won the game if it had been scored as dead. I don't know what
Indigo thought but the jury decided to give the win to Indigo. See
http://computer-go.org/pipermail/computer-go/2005-January/002508.html
for more information about the game, including the game record.
Yes, this one does look like a real dispute.

Also, I don't understand the jury's decision. I haven't listened to the two sides' arguments, I have just looked at the game record. I don't know how I would rule, but on the evidence I have, I don't think it would be for a white win.

If I stand by what I have said so far, and a position like this appears at the end of a game in a KGS tournament:
(1.) If both bots have implemented the game-end protocol as recommended, this can't happen.
(2.) If only one bot has implemented the protocol, I will give the win to that bot, even if it thinks it is its own group in the bottom left that is dead.
(3.) If neither bot has implemented the protocol, I have a headache.

I am not quite persuaded to insist on the protocol being implemented by the next tournament, which will be on June 5th. I feel I ought to give programmers a few months' notice before I start insisting on them supporting the protocol.

Nick
--
Nick Wedd nick@xxxxxxxxxxxxxxxxx
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/