This testing would be ideal for a boinc-style network of cooperative computers.
It isn't time-critical, and does not require a lot of information exchange;
each computer would run a series of multi-way tournaments, and the results
would be collected.
I'd be happy to volunteer part of my desktop computer for this purpose, and
this study might gather as much support as the scalability studies a few months
back. I like the way various tweaks are being compared, to determine their cost
versus value.
The keys to make this work are: low maintenance - a bit of setup and then
forget about it - and some form of feedback to encourage participation.
Terry McIntyre <[EMAIL PROTECTED]>
-- Libertarians Do It With Consent!
----- Original Message ----
> From: Don Dailey <[EMAIL PROTECTED]>
> To: Jason House <[EMAIL PROTECTED]>
> Cc: computer-go <computer-go@computer-go.org>
> Sent: Tuesday, October 28, 2008 12:51:53 PM
> Subject: Re: [computer-go] enhanced test
>
> On Tue, 2008-10-28 at 15:34 -0400, Jason House wrote:
> > You should also test your 7/8 keep heuristic to see if it's reducing
> > weight or avoiding the final moves that matters.
>
> I'm now also testing the reference bot that does NOT check for already
> played moves and it's at least 100 ELO weaker than even the reference
> bot.
>
> So it appears in all cases that checking for "already played" moves is
> quite important.
>
> Rank Name Elo + - games score oppo. draws
> 1 mwNoDup-2000 2106 23 22 578 64% 2003 0%
> 2 mwRefbot-2000 2044 22 22 580 52% 2030 0%
> 3 refbot-2000 2000 22 22 581 43% 2050 0%
> 4 simple-2000 1858 41 44 199 25% 2050 0%
>
>
> These are 9x9, 7.5 komi round robin format.
>
> simple - reference bot without check for already played moves.
> mwNoDup - does the test for already played moves.
>
>
>
>
>
>
> - Don
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
