| |
BGonline.org Forums
GNUbg 0-ply vs. 2-ply rollout studies
Posted By: ah_clem In Response To: GNUbg 0-ply vs. 2-ply rollout studies (leobueno)
Date: Tuesday, 10 May 2011, at 6:40 p.m.
I think it depends on what you're trying to accomplish. If you're trying to distinguish between plays that are very close (say within .02 or so) then gnu 0-ply probably lacks the precision to make that distinction. If you're at a level of play where you're still trying to expunge triple-whoppers from your play, then gnu 0-ply rollout is probably sufficient - does it really matter if your play was a .255 error instead of a .213 error? It's still a blunder, and that's all you need to know.
Most of the time I simply take gnu 2-ply analysis at face value and not bother with a rollout at all since I'm scanning my matches for blunders. I'm not going to worry about the .02 errors until I can manage to stop making blunders, and for this 2-ply analysis is good enough for my purposes - if I could play exactly like gnu 2-ply analysis then I'd be a much better player than I am. If I'm curious about a play, I'll run it through 0-ply rollout, and if that agrees with the analysis I tend to accept that, but if analysis and rollout disagree then it's time for a more thorough rollout.
So, my take is that for Beginner problems (BQ), World-class level computer simulation isn't necessary. If the plays are close, then just knowing that they are close is good enough for beginners, and if they are not close then gnu 0-ply rollout should be sufficient. And I'd rather see a 0-ply rollout than no rollout at all.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.