Re: [Bug-gnubg] My Ideas for move filters

[Joseph Heled]

(Personal opinion)
Maybe i should be more precise. in Analysis, movefilters should work 
intelligently with the user choice of number of moves to show. If she 
wants the top 4 moves, I think we should not filter it down to less.

In fibs2html I simply take 2*N top moves (as given by 0ply) and do a 
2ply for all (N - number of moves asked for). I think analysis in X-ply 
should rely as little as possible on lower plies, and filtering may 
limit the ability of X-ply to find the best moves.

-Joseph


Joern Thyssen wrote:
> On Fri, Jan 03, 2003 at 09:27:18AM +1300, Joseph Heled wrote
>
> > Again, I think the move filters is not applicable to analyze, and I am 
> > not sure about hint.
>
>
> I'm not sure I understand why :-) Can you enlighten me, please?!
>
> I think a move filter with accept=2 (plus a possible number of extra
> moves) at the top level is suitable for analysis/hint.
>
> You don't want to spend your time evaluating 8 moves on 2-ply if only 2
> are really interesting. 
>
> What's your algorithm in fibs2html? Many algorithms can probably be
> expressed in movefilters!
>
> Jørn