[freeciv-ai] Re: omniscience problem

[Jason Dorje Short <vze49r5w@xxxxxxxxxxx>]

Christian Knoke wrote:
> On Tue, Apr 08, 2003 at 06:12:29PM -0500, Mike Kaufman wrote:
>
> > On Tue, Apr 08, 2003 at 09:14:34PM +0000, Per I. Mathisen wrote:
> >
> > > The AI is omniscient and makes decisions based on this. This includes
> > > where to build cities. However, workers, even an AI's workers, cannot be
> > > placed on unknown territory. This presents a problem. Sometimes the AI
> > > places a city on a really great spot, but since much of it is unknown, it
> > > stagnates. An entire AI nation can grind to a halt this way.
> > >
> > > One obvious solution is to allow the AI to place workers on unknown tiles.
> > > This solution may break clients watching the AI, however. Sure would wreak
> > > havoc on CMA.
> > >
> > > I am tempted to go for a more radical solution. Now that we have /away, we
> > > can reveal the entire map for omniscient AIs. This way those of us who
> > > watch AIs, can see what AIs see, which is very useful.
> > NOOOOOOOOO! This  is  the  wrong  direction....
> This is what I thought but didn't say anything.
I disagree; I think this is the proper direction to go to encapsulate 
the handicaps.  Doing it this way is a start toward separating the 
omiscience handicap from the AI code.  So you could, for instance, give 
this handicap to a client-side AI, or to a human (if that code was in 
place).  The big step, of course, is still to change the AI code so that 
it doesn't look at unknown tiles.
To put it a different way, the handicap should be a part of the server 
code, not the AI code.
jason