BGonline.org Forums

Rollout, plus evals of variants

Posted By: Timothy Chow
Date: Sunday, 11 October 2009, at 11:58 p.m.

In Response To: My latest blunder (Timothy Chow)

Seems like my quiz-setting skills are stronger than my backgammon-playing skills! Since I know that you know that I know that you know this is a quiz...

Anyway, over the board, I hit without too much thought. I figured that with equal boards and an advanced anchor, getting hit back wasn't a disaster, and not getting hit back stood to gain a lot. But GNU begged to differ. (Commentary continues after the rollout—scroll down.)

gnubg	139
tchow	137
Position ID: zG4jIBice8AABw Match ID: cAkRAAAAGAAA

#		Ply	Move	Equity
•	1	R	5/1 3/1	-0.095
			0.464 0.126 0.003 - 0.536 0.122 0.005 -0.070 -0.095 0.001 0.002 0.000 - 0.001 0.002 0.000 0.004 0.009
			Full cubeful rollout with var.redn.
			1296 games, Mersenne Twister dice gen. with seed 871312403 and quasi-random dice
			Play: supremo 2-ply cubeful prune [world class]
			keep the first 0 0-ply moves and up to 16 more moves within equity 0.32
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	2	R	13/7*	-0.147 ( -0.052)
			0.458 0.138 0.006 - 0.542 0.144 0.005 -0.091 -0.147 0.002 0.002 0.000 - 0.002 0.002 0.000 0.004 0.011
			Full cubeful rollout with var.redn.
			1296 games, Mersenne Twister dice gen. with seed 871312403 and quasi-random dice
			Play: supremo 2-ply cubeful prune [world class]
			keep the first 0 0-ply moves and up to 16 more moves within equity 0.32
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]

O.K., so what do I make of this? Is this simply the sort of position where hitting is wrong? Suppose we change the position of one of Blue's checkers:

gnubg	139
tchow	139
Position Id: zG4jIBjMe8AABw Match Id: cAkRAAAAAAAA

Now GNU 4-ply says that hitting 13/7* (-0.055) is clearly better than 5/3 5/1 (-0.192). O.K., but maybe the explanation is that it's simply too costly to dump a blot on the 1-point without making a point there. Let's check by trying this position:

gnubg	139
tchow	135
Position Id: zG4jIBice2AABw Match Id: cIkQAAAAAAAA

Here, GNU 4-ply prefers 6/1 (-0.125) to 12/7* (-0.155). Hmmm. So it seems that the exact positioning of the spares is really critical. Moving Blue's outfield point one point closer and changing the roll to 31 confirms this; GNU 4-ply then prefers 11/7* (-0.121) to 5/1 (-0.157).

Finally, it also makes a difference where White's extra blot is. Put it in direct shot range and GNU 4-ply slightly prefers 13/7* (-0.001) to 5/1 3/1 (-0.019), presumably for the tempo.

gnubg	138
tchow	137
Position Id: zG4TIBice8AABw Match Id: cAkRAAAAAAAA

Messages In This Thread

• My latest blunder
  Timothy Chow -- Saturday, 10 October 2009, at 4:27 p.m.
  • My latest blunder
    Daniel Murphy -- Saturday, 10 October 2009, at 6:10 p.m.
    • My latest blunder
      Christian Munk-Christensen -- Sunday, 11 October 2009, at 7:43 a.m.
    • My latest blunder
      alex21 -- Sunday, 11 October 2009, at 5:04 p.m.
  • My latest blunder
    Martin -- Saturday, 10 October 2009, at 7:02 p.m.
  • My latest blunder
    Stick -- Sunday, 11 October 2009, at 1:33 a.m.
  • My latest blunder
    neilkaz -- Sunday, 11 October 2009, at 5:10 a.m.
  • My latest blunder
    Keene -- Sunday, 11 October 2009, at 5:30 a.m.
  • My latest blunder
    jdg -- Sunday, 11 October 2009, at 12:45 p.m.
  • Rollout, plus evals of variants
    Timothy Chow -- Sunday, 11 October 2009, at 11:58 p.m.
    • Rollout, plus evals of variants
      Daniel Murphy -- Monday, 12 October 2009, at 12:25 a.m.
      • Rollout, plus evals of variants
        Daniel Murphy -- Monday, 12 October 2009, at 12:26 a.m.
    • Rollout, plus evals of variants
      Daniel Murphy -- Monday, 12 October 2009, at 12:55 a.m.
      • Rollout, plus evals of variants
        Timothy Chow -- Monday, 12 October 2009, at 9:11 p.m.
    • Rollout, plus evals of variants
      Rich Munitz -- Monday, 12 October 2009, at 3:11 a.m.