| |
BGonline.org Forums
Opening 32-g — Conflicting GnuBg nacbracs — Clarification requested
Posted By: Taper_Mike In Response To: Opening 32: examples of Z outperforming D (Nack Ballard)
Date: Thursday, 19 January 2012, at 8:58 p.m.
Nack does a wonderful job of tabulating and posting the nacbracs for opening sequences. His work enhances the value of the rollouts at BGO. In the case of 32-g, however, Nack posted conflicting nacbracs for GNU Backgammon rollouts of an opening 32 at gammon-go. 2010-Oct-15 — g[D S27 Z29] /15*10
2011-Jan-15 — g[D Z29 S27] /15*10
2011-Feb-06 — g[D S27 Z29] /15*10His report seems to be an aggregate of two GnuBg rollouts, the one here at BGO and another from an unknown source. The BGO rollout is g[D Z28.3 S28.8] /10 5 10, made using the seed 1046052160. No doubt the confusion arises because S, the third-ranked play in the BGO rollout, was given fewer trials than Z, the second-ranked play. I have not yet found the source of the other rollout in the aggregate, but it must be a 5k rollout of DSZ.
Ever the minimalist, I know that Nack prefers to order the plays in a nacbrac so that no trial count need ever be repeated. In order to achieve this, Nack is sometimes forced to list the plays out of order. That is, the moves are not listed in order of decreasing equity. My preference is the reverse. I am less likely to be confused when the plays are strictly ordered from best to worst. The price for my preference is a few extra keystrokes in a nacbrac. Hence the trial counts of "/10 5 10" in my nacbrac for the BGO rollout above.
[Nack, has your minimalism cause you to trick yourself this time?]
Any clarification concerning the trial counts of this rollout of 32-g is appreciated. Is it an aggregate? Do you know the source of the missing GnuBg rollout? And so on.
Thanks.
Mike
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.