Re: [Bug-gnubg] First move in hint list not the best one

[Carsten Wenderdel]

Hi Øystein, thanks for your response!


> Ooooohhhhhh..... Sounds like a bad way to connect it to a bot? I
> guess the external interface or the python interface would be better
> ways to make a bot. Or maybe even Joseph's pygnubg.

Yes, I have to do a lot of parsing stuff, right, but I already got it going.
You are welcome to see it in action, just drop me a line.

I see two problems with the external interface - first it lacks dokumenation
and secondly it lacks features.
I not only need the best move for a certain position but also the equity
loss for a certain move. Therefore I either need the hint list or the tutor
mode, which both are not available through the external interface. Please
correct me if I am wrong.


> Let me make a guess. You parse the list and you find the move with
> the highest equity. The problem is that the list is not necessarily
> sorted with the highest equity move first. If the movefilter have
> sorted out a move at 0-ply and then 1-ply evaluates the candiadates
> with a lower equity, the 1-ply best moves will be sorted hiest on the
> list.

You are right, Øystein.


> In short: The move list is first sorted be evaluation type and then
> sorted by equity. Is that the problem?

The sorting order is not a problem. My problem is that comparing equities in
0 ply and 1ply/2ply is like comparing apples and oranges.
So if I want to compare it correctly it seems I have to use 0 ply. Or is
there any way to computate a 2 ply equity for _every_ possible move?
Another possibility would be the tutor mode ... hm.


Anyway, your answer helped me much, thanks!

-- 
DSL Komplett von GMX +++ Supergünstig und stressfrei einsteigen!
AKTION "Kein Einrichtungspreis" nutzen: http://www.gmx.net/de/go/dsl