Re: About all the AI
On Thu, 3 Jul 1997, Ralf Nieuwenhuijsen wrote:
> Maybe i am wrong, but aren't you guys loosing your objective
> here....
I'm not. My objective ia clear. He asked "how do I create AI?" and I
told him. It's great to get other ideas from the crowd.
> It is not about beating the other about who's better.....
> But to create the best AI possible for Joseph Martin's Risk game.
There is no "best AI possible." Everyone who plays against the AI will
have a different opinion about it's strengths and weaknesses. What most
people (who haven't written an AI before) would call a great AI just plain
isn't. We have to stop thinking in terms of right and wrong. We need to
think about what's fun.
> Also you both say about the same thing implended a bit differently,
> i looked and thought about your explenations for about an hour, cause
> they were interesting and i dunno anything about AI.
> I noticed that Packard's idea is more general for AI, Anders way is
> also included in his method, on the other way.... Anders method is
> more practical and when you have written the code you can use the
> code again easily....
Our ways are identical in practice. The only difference is in the
decision tables that drive the AI. Why do I do something, when do I do
it, when not to do it, when do I panic?
Usually game AI code is game specific, rarely do you re-use major portions
of the AI code in a different game. Most of BattleChess' AI is worthless
in Duke Nukem 3D.
> You should begin a contest.... see which AI is the best and then
> combine it and sent it to Joseph Martin for his game....
I'll advise, I'll playtest, I'll give opinions. He can do whatever he
wants with them...
Michael Packard
Lord Generic Productions
lgp at exo.com http://exo.com/~lgp
A Crash Course in Game Design and Production
http://exo.com/~lgp/euphoria
|
Not Categorized, Please Help
|
|
