Re: [Bug-gnubg] Predefined move filters

[Joseph Heled]

Albert Silver wrote:
> I'll do some testing of this, but it's not always easy to see the
> difference. 

I would say almost impossible without use of something like the rollout 
benchmark. I never do any work in 3ply, so I have little idea what a 
good filter will be at this level, but it seems reasonable to assume it 
does not suffer from the 1ply syndrome, since it "sees" much more of the 
future.

Perhaps later I will try the 3ply with the benchmark, and only try to 
get some numbers on various 2-ply filters.

-Joseph

One detail: It's true that the 1-ply is the worst (IMHO) ply
> depth, however, in my experience, the 3-ply is definitely stronger than
> the 2-ply. This has been confirmed by numerous rollouts, as well as
> errors pointed out by Snowie. As a rule its evaluation is more precise,
> and it is able to find the correct moves many times that it might miss
> at 2 plies. This includes its chronic problems with bearoff against more
> than 1 checker on the bar. 
>
>