[freeciv-ai] Re: two major tactical ai weaknesses

[Christian Knoke <chrisk@xxxxxxxxx>]

On Mon, Feb 21, 2005 at 10:57:14AM -0500, Benoit Hudson wrote:
> On Mon, Feb 21, 2005 at 01:06:14PM +0100, Christian Knoke wrote:
> > 
> > Hello,
> > 
> > I just want to note something you probably already know. Some AI weaknesses
> > which - if absent - could improve AI a lot (maybe too much ;-) ):
> > 
> > - AI diplomat(s) buys cities - and then goes to the next city, the just won
> > city remains empty.
> > 
> > - AI stacks attack cities successfully - but then do not conquer them.
> 
> I've seen this one.  Typically, the AI attacks and wins but the unit is
> badly damaged.  It seems that self-preservation (fleeing to the nearest
> allied town) overcomes the lure of a free city.

Not in the case I've just seen. 7 alpine troops move in front of my city;
*one* of them attacks and wins, the city (size 5 or so) is empty, and no one
moves in. A massive attack would have been necessary to get them out again.
The only thing that stood against was they had no diplomat with them - but
then, this is a systematic problem, to have a diplo in the stack from the
beginning of the attack.

> It's not entirely crazy: you don't necessarily want to take a city just
> to lose it again.  But it's usually wrong.

Christian

-- 
Christian Knoke            * * *            http://cknoke.de
* * * * * * * * *  Ceterum censeo Microsoft esse dividendum.