| |
BGonline.org Forums
Measuring Complexity
Posted By: Phil Simborg In Response To: Measuring Complexity (Michael Depreli)
Date: Thursday, 3 September 2009, at 4:39 p.m.
In Woolsey and Heinrich's book, New Ideas In Backgammon, they do a very good job of tackling this very question in the selection of the plays. They explained their selection process very completely in their introduction, but it included choosing positions that they found difficult; choosing positions that the bots found difficult and had to roll out to get confidence; and putting positions in front of world class players to see how they fared with them.
I have developed a combination test/lesson showing 12 moves and I give scores of anywhere from 5 points to 10 points for getting the best move. I award 10 points if the right move is "difficult" to make, and 5 if the right move is relatively simple.
To determine what rates as difficult and what rates as simple, I used the same approach that Kit and Hal used, but I also looked at another area. If the difference between the best play and the second best play was not great, but the two plays represented completely different approaches (e.g. hitting vs. priming; going for the G or playing safe), then I considered this play to be more "important" to get right, and therefore award more points for it.
Bottom line is it will always be somewhat arbitrary and virtually impossible to do on a completely analytical basis (in my opinion, of course).
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.