On Tue, Aug 4, 2009 at 9:44 PM, Michael Petch
<address@hidden> wrote:
Howdy Team,
I find that a lot of the fringe cases (with such incorrect results) like
this usually occur on the race evaluator near end games. If this is a result
of pruning, maybe we might consider not pruning when the race evaluator is
being used?
The position showed, the backgammon rate is evaluated by a code called "race improvements". It takes the distribution of rolls to save bg and compares it to the distribution of number of rolls to get off for the other player.
I have never seen any bugs in the "race improvement" code. Maybe the code is not used together with "pruning" and the correct move is pruned away (at some level) for that reason. Or, maybe the code evaluates differently depending on -O flag level and system?
I have not browsed the code on this issue.
-Øystein