Hi Josh, Here's upstream's response to your bug report. The summary is that this behavior is intentional, and upstream doesn't believe that any significance should be placed on the final score of a game.
In some subsequent messages, Michael pointed out that there are several other places where the gnubg evaluation algorithm will lead to unintuitive play to a human because the computer knows that all options are identical when it comes to winning or losing the match and will essentially pick one of the options at random. The other major example was that the computer won't put any weight on avoiding a gammon if the opponent is one point away from winning a match. Eliminating all of these cases from the engine would apparently be quite a lot of work and isn't something that he thinks is worth spending any time on. Michael Petch <mpe...@capp-sysware.com> writes: > This is my opinion only (others may wish to chime in), and I was part of > the discussion on this list previously. I consider the behavior > intentional. Resigning more points than necessary (albeit correct) still > is a win. How many points one wins by isn't relevant. Resigning fewer > points is an issue, resigning more at the end of a match doesn't make > any difference. > Having an option for this is pointless. GNUBG has a lot of options as > is, but adding one that has no inherent value, and seems (IMHO) > absolutely pointless to me. > I say stick with what exists, and I consider this complaint invalid from > a user. A user has to understand that in match play, a win is a win no > matter how many points it is by. That is what I would tell the end user. > Tom Keith's www.bkgm.com is a source of everything Backgammon (almost > like a online bible for the game). Have a look at: > http://www.bkgm.com/rules/match.html . There is this paragraph that sums > it up nicely: "There is no bonus for winning more than the required > number of points in match play. The sole goal is to win the match, and > the size of the victory doesn't matter." . Given the response, my inclination is to close this as wontfix, since I think it's unlikely anyone will ever put the work into changing the behavior. -- Russ Allbery (r...@stanford.edu) <http://www.eyrie.org/~eagle/> -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
