On May 21, 2009, at 5:28 PM, Don Dailey <dailey....@gmail.com> wrote:
On Thu, May 21, 2009 at 5:09 PM, Michael Alford <m...@aracnet.com>
wrote:
Don Dailey wrote:
I believe with CGOS rules, it is believed to be 9.0
I don't know if there is a proof of that, but I don't think there is
any dispute.
- Don
On Thu, May 21, 2009 at 4:29 PM, Michael Williams <michaelwilliam...@gmail.com
<mailto:michaelwilliam...@gmail.com>> wrote:
What was the consensus on 7x7 komi? It was discussed back during
Don's scalability study, but I couldn't find the number itself.
Was
it 9.0?
_______________________________________________
computer-go mailing list
computer-go@computer-go.org <mailto:computer-go@computer-go.org>
http://www.computer-go.org/mailman/listinfo/computer-go/
---
---------------------------------------------------------------------
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
delurk/
9 may be too small, the smaller the board, the more komi w needs,
however, when i teach beginners, the more interesting variation is
for w to try and live in two places :)
I feel very confident that 9 is not too small. With 9.5 komi
white wins every game.
We did experiments with go programs on 7x7 and with 9.5 komi white
won almost every game. When the komi was reduced to 8.5, black won
nearly every game.
If you think 9 is too low, you can probably repeat this
experiement. Try beating one of the strong MCTS programs on 7x7 as
black with 9.5 komi. I'll bet you cannot do it.
7x7 isn't solved by computer, but the best ones play it extrememly
well. Does anyone have any information on how well they play it?
My guess is that with 9.5 komi, a strong computer playing white
won't lose much to anyone (as it's starting from a dead won position.)
There's also an SGF file with analysis of variations and results done
by some amateur dans. It also concluded the correct Komi is 9_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/