[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

64-g ^15 RO

Posted By: Daniel Murphy
Date: Friday, 25 February 2011, at 4:17 p.m.

In Response To: 64-g ^15 RO (Stick)

Stick's hit all the main points.

I might add this: Suppose you were going to roll out, say, the top three plays for all opening rolls with Gnubg 0-ply/0-ply, 0-ply/2-ply, 2-ply/2-ply, and 4-ply/4-ply. If you hypothesize that most of the time all rollouts will rank the three plays the same, I'd be very surprised if your hypothesis did not turn out to be correct. 0-ply has "world class" strength, after all. But how would you predict in advance which low-level rollouts would not match the ones with better parameters? If you didn't have the better rollouts, how would you know which of the low-level rollouts were correct?

And then, as Stick mentions, play rank may not be the only result we're interested in. We may want a better idea of absolute equities and absolute differences between plays.

I have different standards for rollouts I do for myself, and rollouts I do to share. For quick review for my own use only, I may be satisfied with ballpark accuracy, not much concerned with absolute equities, and more willing to trade certainty for speed. So, I've often used 0-ply checker play. But not 0-ply cube; it's not so "world class."

There's a couple or three more things to be said about 0-ply vs. 2-ply accuracy and speed, which Chuck Bower said here:

http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=23735

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.