Re: [gnugo-devel] 261 games 3.1.31 vs 3.1.32

[bump]

> > As I don't think the commercial versions of the top programs include GMP
> > (not to speak of GTP), you cannot do such a series automatically.
> 
> Yes, it would take time. On the other hand Many Faces is playing on IGS 
> and/or NNGS, but it is probably not appropriate to tie it to play 
> regression games against GNU Go...

There are a couple of games with Many Faces in the regressions that
were obtained on NNGS. I think for copyright reasons we'd not use
games played on IGS but I doubt if David Fotland minds if we play
a game once in a while when he shows up on NNGS.

> On Friday, April 19, 2002, at 04:31  pm, Daniel Bump wrote:
> > About the strangeness of the games, current GNU Go will often start
> > to rescue a group, then abandon it because something else seems
> > bigger. Sometimes this works out OK because the opponent may invest
> > several moves to kill the stones in question. On the other hand
> > inconsistent play is a problem that needs to be fixed.
> 
> By strange, I was thinking about cases where both GNU Go instances seem 
> to avoid contacts and fights and make many extensions (ikken tobi, 
> kogeima, kosumi...). This sometimes result, after 50 or so moves, in a 
> large spread of "loose" stones all over the board, with patterns that 
> seem unnatural to me, or at least very different from what I see in 
> human games. Am I making sense? And if so, this is probably to be 

I think we'd have to discuss particular examples. 

> expected, but then I would think that it shows that playing GNU Go 
> against itself is not a very good measure of its progress or strength.

I agree that although GNU Go 3.1.32 can give GNU Go 3.0 a 3 stone handicap it
is not 3 stones stronger against humans. On the other hand it has not been
explicitly tuned against 3.0, so the benchmark is not completely
meaningless. I think the increase in strength over 3.0 is about 2 stones based
on results on NNGS.

Dan