RE: [Bug-gnubg] Multi-bot rollouts of backgame position (Joseph Heled

[Albert Silver]

I never thought using 1-ply filters when analyzing 2-plies to be good. I have gotten skewered results and though 1-ply may actually be a fraction better than 0-ply, I don’t think it’s enough to justify cutting down on the number of moves analyzed at 2-ply. In fact the setting used seems only a little better than World Class as it has a wider equity filter but only analyzes 7 moves at 2-ply. I still think the best compromise is the one I use now being no 1-ply filters and 0.320/12. It’s about 25% faster than Supremo and pretty much as efficient as far as I’ve seen until now. Just MHO.

 

Albert

 

-----Original Message-----
From: address@hidden [mailto:address@hidden On Behalf Of address@hidden
Sent: Sunday, March 09, 2003 2:01 PM
To: address@hidden
Subject: [Bug-gnubg] Multi-bot rollouts of backgame position (Joseph Heled

 

In a message dated 09/03/2003 12:12:08 GMT Standard Time, address@hidden writes:

I don't buy it.

If 2ply came out worse than 0ply I say:
   Either
     He is using the wrong filters for 2ply.
     He is using the version before the rollout bug fix.
     The dice was different for the rolls, with particularly bad dice
for the 2 ply.

-Joseph

I used 0.308/13 and 0.154/7 for the filters to match Snowie's
Are these "wrong" ?
I used 03March03 build

It's possible of course that 0 ply doesn't bear off as well against the anchor(s)
and loses equity making it look as if the backgame is being played better.