BGonline.org Forums

15552-game rollout of 41Z-65R-43

Posted By: Rich Munitz
Date: Sunday, 27 December 2009, at 7:12 p.m.

In Response To: 15552-game rollout of 41Z-65R-43 (Timothy Chow)

I agree. I think that truncated rollouts are too easily dismissed around here for cases where only a limited amount of time is to be invested in the rollout - if and only if the bot has a decent understanding of the absolute equity of the position type. Its absolute equity evaluation errors should balance out to produce a rather good average.

However for situations like openings where someone is going to run the rollout to statistical significance no matter how long it takes (meaning that computation time is not a concern), then it is clear to me that full rollouts are superior.

When the absolute truth must be known, full rollout is the only way to go. But absolute truth is not always the goal in every situation.

Lets keep in mind that XGR or XGR+ which is a pretty awesome backgammon player is nothing more than a truncated rollout, and for things like evaluating match play, though not perfect, I think we can all agree that it is a far more efficient use of computer resources (bang for the buck) than trying to do full rollouts on all mistakes to the point of statsig.

Messages In This Thread

• 15552-game rollout of 41Z-65R-43
  Timothy Chow -- Sunday, 27 December 2009, at 12:23 a.m.
  • 15552-game rollout of 41Z-65R-43
    Leonardo Jerkovic -- Sunday, 27 December 2009, at 12:34 a.m.
  • 15552-game rollout of 41Z-65R-43
    Stick -- Sunday, 27 December 2009, at 12:41 a.m.
    • 15552-game rollout of 41Z-65R-43
      neilkaz -- Sunday, 27 December 2009, at 1:03 a.m.
    • 15552-game rollout of 41Z-65R-43
      Timothy Chow -- Sunday, 27 December 2009, at 2:08 a.m.
  • 15552-game rollout of 41Z-65R-43
    Nack Ballard -- Sunday, 27 December 2009, at 1:38 a.m.
    • 15552-game rollout of 41Z-65R-43
      Timothy Chow -- Sunday, 27 December 2009, at 2:02 a.m.
      • there are lies,
        garyo -- Sunday, 27 December 2009, at 4:00 a.m.
        • rollouts make an assumption
          garyo -- Monday, 28 December 2009, at 3:09 a.m.
      • 15552-game rollout of 41Z-65R-43
        Chuck Bower -- Sunday, 27 December 2009, at 2:52 p.m.
        • 15552-game rollout of 41Z-65R-43
          Timothy Chow -- Sunday, 27 December 2009, at 6:47 p.m.
          • 15552-game rollout of 41Z-65R-43
            Rich Munitz -- Sunday, 27 December 2009, at 7:12 p.m.
          • 15552-game rollout of 41Z-65R-43
            Chuck Bower -- Sunday, 27 December 2009, at 7:25 p.m.
          • 15552-game rollout of 41Z-65R-43
            Chuck Bower -- Monday, 28 December 2009, at 2:14 p.m.
            • 15552-game rollout of 41Z-65R-43
              Timothy Chow -- Monday, 28 December 2009, at 7:33 p.m.
              • 15552-game rollout of 41Z-65R-43
                Maik Stiebler -- Tuesday, 29 December 2009, at 8:45 a.m.
          • 15552-game rollout of 41Z-65R-43
            Daniel Murphy -- Wednesday, 16 June 2010, at 1:57 a.m.
        • 15552-game rollout of 41Z-65R-43
          Tom Keith -- Sunday, 27 December 2009, at 6:57 p.m.
    • 15552-game rollout of 41Z-65R-43
      Daniel Murphy -- Tuesday, 29 December 2009, at 12:58 a.m.
    • 41S-65R-43
      Matt Cohn-Geier -- Wednesday, 16 June 2010, at 1:06 a.m.
      • 41S-65R-43
        Nack Ballard -- Wednesday, 16 June 2010, at 1:53 a.m.
        • 41S-65R-43
          Stick -- Wednesday, 16 June 2010, at 3:30 a.m.