| |
BGonline.org Forums
Real Backgammon
Posted By: Daniel Murphy In Response To: Real Backgammon (Bill Riles)
Date: Monday, 26 April 2010, at 9:48 p.m.
Bill: "In my estimation, too many who are overly concerned with minor errors in opening play seem to view a backgammon game, money or match play, as some type of trajectory or trigonometry problem -- ..."
My view is that against an even opponent I rate to be 50/50, but if I play against an otherwise even opponent who has mastered opening play while I have not, I've changed my odds slightly, say, to 48/52. Probably not much worse, and possibly not that bad, but why concede a slight edge, when opening play errors are so easily studied and corrected?
Before taking up serious competition again, I'd certainly devote many hours time to brushing up on opening theory. My goal would not be to avoid making a single tiny according-to-bot error. My goal would be to be able to play my choice of 1st and 2nd moves with hardly a thought at all scores, understand the dynamics of the resulting positions, and internalize an efficient and rational process for judging nonobvious 3rd and 4th moves and cube decisions. I'd probably dredge up Chuck Bower's old Flint articles on second roll responses based on his JellyFish rollouts, compare them to more recent bot results, and review the simple and useful heuristics Chuck gave for recognizing best plays over the board without having memorized all possible continuations. I'd probably also compare the older rollouts at Tom Keith's site with the newer ones here and elsewhere, because the older rollouts influenced my thinking some years ago, and I'd want to reevaluate what I think I know. Another thing I'd do is something I've found helpful before, which is play sessions against a bot restarting after 5 moves, evaluating, and thinking about the errors the bot has flagged (some of which aren't errors). In my experience, having resumed such study after an extended period of nonactivity, I can rid my game of half its 1st-5th move errors with just a few hours of practice.
It's true that many rollouts of opening plays (1st-4th moves) show no significant difference between two or more plays. As Bill wrote earlier: "If two moves are that close does it really matter which you make? Go with the one that makes you most comfortable ...." I agree. But that is, actually, just what such nearly-identical-to-four-decimal-places rollouts are telling us.
I agree, too, that overly focusing on inconsequential opening play decisions is unproductive (that's overly obvious!). But other rollouts -- I think David Rockwell posted some examples several months ago -- show that the difference between two plays that may both seem reasonable to an advanced or expert player may actually be quite large. I don't think it's unlikely that someone who little time to studying openings might frequently be making 0.050 errors or worse in the first few moves.
"... where an initial minor variation results in a huge error over the length of the course. This, obviously, is not true."
I'm not sure why that is obvious. If I were super-expertly making only a tiny 0.010 error in the first five moves of each game of a 10 game session, would it be wrong to think that I have made the equivalent of a 0.100 blunder over the course of the session? And I think it's likey that most folks who haven't given much thought to correcting their opening errors, or a rusty non-super-expert such as myself, is very likely to be giving up a whole lot more than that.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.