Complete.Org: Mailing Lists: Archives: freeciv-dev: November 1998:
[Freeciv-Dev] AI weaknesses + idea for player handicaps (was: What impro
Home

[Freeciv-Dev] AI weaknesses + idea for player handicaps (was: What impro

[Top] [All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index] [Thread Index]
To: freeciv-dev@xxxxxxxxxxx (Freeciv developers)
Subject: [Freeciv-Dev] AI weaknesses + idea for player handicaps (was: What improvement to build next...)
From: Reinier Post <rp@xxxxxxxxxx>
Date: Tue, 3 Nov 1998 19:43:57 +0100

> Perhaps adding personalities to the AI would help answer a lot of these
> problems?  It may not be that the AI is particularly good - just that it
> goes straight for the kill without holding back.  It's like constantly
> playing the Zulus, except much less stupid with regard to tech.

Mmm, I rather think the AI is overly defensive.  Occasionally it sends
out units into enemy territory, but these units are mostly defensive
troops (alpines, not cavalry; mech infs, not howitzer) and they rarely
attack.  AI players do not actively attack their opponents, they have
no sense of strategy.  But tactically, they are reasonably good.  The
exception are cities they consider to be within its own zone of
influence, these can expect a wildfire of attacking units.  Apart from
that, local tactical manoeuvering is all that ever happens.

> Making the AI (or some AI players) concentrate less on conquest and more
> on developing the civilization should allow new players to get their
> feet more easily.

But this is exactly what the AI do!  Until they are out of unexplored
territory; from then on, they start to defend.  An AI player can only
win a game by exhausting its opponents: if an opponent is weak enough,
finally all the random tactical manoeuvering will reduce him/her to zero.
Try observing an all-AI game with generator 1 and a large landmass.

The biggest strength of AI is not its attacking power, I think,
it's its incredible growing power in the expansion stage.

The cost of building improvements and units is currently fixed with
each type.  This cost could be varied with players.  A 'player profile'
is just a stored vector of costs; it is a simple way to represent
different personalities and skill levels.  It also allows us to
examine the effect of different building strategies, by pairing off AI
players with different profiles.  The AI's early expansion can be tempered
by increasing the cost of its settlers; its aggression, by increasing that
of its military.

This may even work well with the existing AI code.  More personality
traits and handicaps could be introduced there, of course.

-- 
Reinier Post                                             rp@xxxxxxxxxx


[Prev in Thread] Current Thread [Next in Thread]