Re: A diplomacy AI rule

Well, 
I agree with Per that the goal is to win.

But to win, a civilization may first think to survive long enough.
And does more complex algorithm (but implementable), and symmetric, could
give more fun to players ?
Could multi-part symmetric alliances, instead of the best against the
others, be more challenging for human players ? Less predictable ?
Perhaps the algorithm should be parametrised (hook functions ?) to enable
different ways to do diplomacy, as banjo said ?

I know that your current task of achieving the 2.0 is challenging, and I
hope that this little debate was fresh air.

Thank you for your time and attention,

Olivier


-----Original Message-----
From: banjo [mailto:[email protected]] 
Sent: mardi 7 décembre 2004 09:48
To: Olivier DAVY
Cc: freeciv-ai-0/[email protected]
Subject: Re: [freeciv-ai] A diplomacy AI rule

Olivier DAVY wrote:
> Hi,
> 
> I got an idea for the diplomacy system of freeciv.
> It will force the IA to use diplomacy (between IA and IA/Humans).
> 
> The following rule could be used :
> 
> The opportunity that diplomacy adds is that a civilization can become 
> the ally of another.
> Being in mind that EVERY civilization (symmetry ?) has its 
> opportunity, what sould they do ?
> It can be seen also has a constraint for IA : if the civilization does 
> not use this opportunity, the other will do, and will crush the former
easily.
> 
> I think (well, after reading the fourth book of "Foundation") that 
> civilizations should tend to create alliances that will create equal 
> forces, i.e. they should create symmetric super-civilizations (the 
> alliances) to compensate the fact that civilizations have not the same 
> power. They hence adjust the forces and live forever.
> What is important is symmetry in that case.
> Can we say that symmetry is the base of peacefull civilizations (e.g. 
> : the Cold War ?).
> 
> Practically, every ten turns (for example), we should compute the 
> forces ratio, and allow IA civilizations to recreate or delete their 
> alliances (but at what cost ?).
> 
> If you find that idea could be interesting, here is an example
> Civ        Force / Points
> A          300
> B          200
> C          101
> D          30
> E          1
> 
> Perhaps an alogrithm could force IA to make alliance such as A-D = 330 
> and B-C-E = 302 ?
> 
> The algorithm could be "try to create #n alliances that minimize the 
> maximum of difference between two of them".
> Thus everybody will think twice before attacking whatever civilization.
> 
> Perhaps a civ can make peace with civ that are very small (to attack 
> them later ?), and create alliance with same-force-civs.
> 
> Since I am also an engineer in computer science, perhaps I could help 
> to go further, if you like the idea.
> 
> Olivier DAVY
> Olivier.davy_at_free.fr
> 
> 
> 
> 
> 


or it could be an option that tunes the ai's level of diplomatic sloppyness,
maybe a way to change the grand strategic game by creating more unstable
alliances, unlike the long lived ones I (we?) play at present.

what happens if we put knobs on some constants to see how they change the
feel of a game.

-banjo