Open Side Menu Go to the Top
Register
Interesting 5-4 in (proto?)backgame Interesting 5-4 in (proto?)backgame

12-13-2010 , 10:08 PM
I don't know how to post images here...so I also posted the problem http://www.bgonline.org/forums/webbb...mes;read=83557

enjoy
Interesting 5-4 in (proto?)backgame Quote
12-13-2010 , 10:18 PM
Well, the 3 moves I was looking at were the 3 you bothered to roll out, so I guess that's a good thing. Choosing them in exactly reverse order, not so much.
Interesting 5-4 in (proto?)backgame Quote
12-14-2010 , 10:09 AM
If you ad both confidence intervals, it is still bigger then the difference of the candidates. I think 3ply is strong enough. I would therefore crank up the trials. Some guys with good machines make 4ply RO's for the openings, but even that will last days. Looking at the triangle confidence interval, accuracy and time consumption, i think we can get beyond 3ply, if we have cuda technology with teraflop performance. But let us look at the suggestions of Dr. neilkaz when XG 2.0 arrives.
Interesting 5-4 in (proto?)backgame Quote
12-14-2010 , 12:46 PM
Double hit, try and build another point or even two in our board and gain control of the outfield imo. Roughly to square one if hit, but in a much better position if we can build another point and keep a man or two in the air and get some of ours moving around.

Obv not ideal to blitz a bit with not enough men but think there is enough here to try and at least buy some time and let us get into a much better position to attack again later.
Interesting 5-4 in (proto?)backgame Quote
12-14-2010 , 04:46 PM
Quote:
Originally Posted by higonefive
If you ad both confidence intervals, it is still bigger then the difference of the candidates. I think 3ply is strong enough. I would therefore crank up the trials. Some guys with good machines make 4ply RO's for the openings, but even that will last days. Looking at the triangle confidence interval, accuracy and time consumption, i think we can get beyond 3ply, if we have cuda technology with teraflop performance. But let us look at the suggestions of Dr. neilkaz when XG 2.0 arrives.
The confidence of the top move vs the other 2 was over 95%
Interesting 5-4 in (proto?)backgame Quote
12-14-2010 , 05:42 PM
Double hit seems to have 2 ways to win (alternatives I considered were 8/3* 7/3 and 8/3* 9/5). You can partially contain a checker and hope he has to play bad dice up front and/or play for outfield hits while you come around, or you can get hit and fix your timing somewhat. Double-slotting is basically begging to get hit and fix timing, and pointing on him is trying to play the outfield plan with only one checker to target instead of 2 (since your timing is questionable at best). Yesterday, before looking at rollout, I wanted to double slot and try for the timed backgame.. but you don't lose a lot by giving yourself a chance to win coming forward as well.
Interesting 5-4 in (proto?)backgame Quote
12-14-2010 , 05:54 PM
I think that sometimes it pays just to make a play in a complex position that you know can't be too far off, instead of making one that can be either the first play or far down the list. Here, I just make the three point on my opponents head. 8/3* 7/3
Interesting 5-4 in (proto?)backgame Quote

      
m