|
Post by Steve Draper on Aug 28, 2014 4:29:48 GMT -8
Thanks for getting this done (Alex I assume?). The results are interesting. At least for Sancho, play quality looks distinctly lower than in Speed Chess, which I think is down to the way the goals are graduated rather than binary (and possibly non-fixed-sum is contributing too). It's an interesting effect.
|
|
|
Post by alandau on Aug 28, 2014 15:42:08 GMT -8
I suspect the non-zero-sum aspect leads to most of the unexpected behavior that's seen. It's much harder to anticipate opponents' behavior in non-zero-sum games, which in turn leads to questionable play based on bad assumptions.
I'm thinking about adding a zero-sum variant, i.e. start at 50/50 goal values and add points based on the difference in the number of pieces, or maybe just grant 100/0 to the player with more pieces left at the end. These would result in more intuitive play, with players trying to protect their own pieces as they capture the opponent's.
|
|