Re: [Bug-gnubg] more on movefilters

[Jim Segrave]

On Fri 13 Dec 2002 (06:45 +1300), Joseph Heled wrote:
> FindnSaveBestMove(Personally, I think The move filetrs belong only in 
> FindBestMovePlied.
> 
> FindnSaveBestMove (which is used for analysis) should use another strategy. I 
> filter down to 2*N moves using 0 ply, then choose the best N with the 
> requested 
> ply. Otherwise I can't see the benefit of the analysis done in 2ply if most 
> of 
> the moves get thrown out by 0ply and have no change of getting evaluated.
> 
> -Joseph
> 

I'm just curious - could you give more of a sketch of how you think
analysis should be done instead of using movefilters? They seem to me
to be ideally suited to the task, but I assume I'm overlooking something.

-- 
Jim Segrave           address@hidden