RE: [computer-go] Pattern matching - example play

["Mark Boon" <tesujisoftware@xxxxxxxxx>]

> -----Original Message-----
> From: computer-go-bounces@xxxxxxxxxxxxxxxxx
> [mailto:computer-go-bounces@xxxxxxxxxxxxxxxxx]On Behalf Of Thore Graepel
> Sent: Sunday, November 28, 2004 0:31
> To: computer-go
> Subject: RE: [computer-go] Pattern matching - example play
>
>
> Mark,
>
> I think your analysis is quite interesting but somewhat lacks the
> "objective" value of Frank's statistics in terms of professional move
> prediction.

When it comes down to just the prediction of pro moves, yes I agree.

There are two things here that keep getting confused. On the one hand
there's the use of Frank's system as a study tool for joseki and fuseki. For
this it's maybe important to have the more "objective" statistics you refer
to. I'm not making any comments in this area as it doesn't interest me that
much.

On the other hand there's the question how useful it would be to have such a
system in a playing program. Frank has been more careful in his more recent
posts about what his system could mean for a playing program than in earlier
posts where at some point he claimed: "A strong Go program will become much
stronger with my system." For someone like me, who is mostly interested in
Go playing programs, that question is a very interesting one.

On the surface, a system that can play 90% of the first 100 moves like a pro
may seem pretty useful to a playing program. Useful, definitely. But more
useful than the joseki-libraries and pattern-databases already in current Go
software? I have sincere doubts about that. As must be clear by now, I have
a strong opinion in that regard. But an opinion by itself doesn't count, one
should always be willing to discuss it and be open to new ideas and
findings. With my previous 'analysis' I try to give a basis to my opinion
where I think a system that can come up with really good moves 90% of the
time is not going to have a big impact on the level of play of Go software
as such. If someone sees a flaw in my analysis, or can somehow show me that
such a system in fact would greatly improve the play of a Go program, then I
would be very interested to hear about it. Previously I think this
discussion has had too much confusion to what it's all about and in some
cases has been taken too personally. That is a pity, as I think it's
actually one of the most interesting subjects in this group.



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