[computer-go] Corrected result

[Peter Drake]

I previously reported an experimental result that sometimes choosing  
the (apparently) best move from a node rather than a random move  
never helped. This sounded awfully fishy, but I couldn't find a bug  
that would explain it.

I've since found it. There was one point in the code for determining  
the best move where I was looking at the children of the root rather  
than the children of the current node. Now it appears that choosing  
the best move around 50% of the time is best.

Peter Drake
Assistant Professor of Computer Science
Lewis & Clark College
http://www.lclark.edu/~drake/




-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://computer-go.org/pipermail/computer-go/attachments/20060902/1e447939/attachment.htm