Re: [computer-go] SlugGo v.s. Many Faces update

["Paul Tabor" <paul@xxxxxxxxx>]

Surely the use of a particular level for komi depends on an assumption that the players value playing first at this level. In recent years 5.5 points (used in both cases quoted recently on this board) has been considered quite low. Rather like the variance in playing strength, I would have thought it reasonable to assume that the fair value for komi may also be different between computers and people, and maybe between different computer programs.

In an even competition, surely there would be value in playing the same number of games 'each way'. Data from this exercise could be used in tuning the komi value.

Obviously this problem only applies to even games and doesn't arise when trying to establish a difference of 1 or more stones between two players.

Paul

----- Original Message -----
From: "Darren Cook" <darren@xxxxxxxxxxxxxxxxx>
To: "computer-go" <computer-go@xxxxxxxxxxxxxxxxx>
Subject: Re: [computer-go] SlugGo v.s. Many Faces update
Date: Fri, 01 Apr 2005 19:12:31 +0900

> 
> > Another question though: it's asserted above that MF is one or two
> > stones stronger than GnuGo. Is this backed up by any data?
> 
> No, David Doshay posted the below in gnugo-devel earlier today.
> 
> Darren
> 
> 
> You may be pleased to hear that GNU Go 3.6 level 10 appears
> to be very evenly matched at this point with MF. In a 67 game
> set:
> 
> B(MF) 32
> W(GG) 35
> 
> Also, white appears to win by a bigger margin more of the time.
> Including one game by 103.5. These results are using komi 5.5.
> _______________________________________________
> computer-go mailing list
> computer-go@xxxxxxxxxxxxxxxxx
> http://www.computer-go.org/mailman/listinfo/computer-go/

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