current boardgame I have.Well it may or may not be more fun than Risk. . . . It is a great game to feature as a mini game. Or perhaps make a huge 2player version.
Anyway, the AI that I always play against is horrible. I am wondering what is involved with creating a hard computer player? (That is, one with a high win %?...or one that is fast. What are the trade offs?)
Some things that I think an agent would need to do:
1) Create strong starting position (for example, don't accidentally trap key pieces) before game has even started.
2) Collect material. This comes in at least two parts:
a)Preserve your pieces. Don't risk killing valuable pieces if you are "not positive" that you will win. (This idea is called "fuzzy logic." Here is a great article from BYOND about using fuzzy logic in your game..
b)Capture when possible. Computing when exactly you should and should not capture may be difficult. It would involve the priority of each piece, and a few other things that are too much right now.
3 Save/analyze past states. Perhaps debatable. . . . Could an unbeatable agent be created if it were allowed to process the entire game? Is that fair for other players/agents?
No comments:
Post a Comment