[Freeciv-Dev] Re: ICS: double empire in 10 turns

[Thomas Strub <ue80@xxxxxxxxxxxxxxxxxxxxx>]

On Thu, Dec 18, 2003 at 03:40:08AM -0800, Brandon J. Van Every wrote:
> James Canete wrote:
> >
> > I decided to do a quick back-of-the-envelope calculation of how fast a
> > "pure" ICS player would grow, given "ideal" conditions.
> 
> It's an interesting calculation.  I'd rate the utility of Freeciv as a
> tool on whether it could *actually quickly test* such calculations.
> Like, rapidly code up an AI with this strategy for spreading over a map,
> and see what the ideal results are.  That sounds hard to do in the
> current codebase however.  Am I mistaken?

It wouldn't be that hard to write an AI that is only good at settling.
But freeciv is more than only settling.

You need:
Exploration (ok AI doesn't, because it is cheating)
Cityplacement (that is that we want to optimize)
Terrain improvement
Research
Attacking/Defense (with building units)
Building Wonders/Cityimprovements.

Because you need all of it, and you have different maps every game. It's
hard to implement an good AI.

When you reduce the number of problems which the AI has to solve the
quality from the AI increases.

> > Anyhoo, how do those sound?  Completely ludicrous or no?
> > I apologize in
> > advance if I'm just yelling out things people have already
> > pointed out.

Your computation isn't 100% correct because in the later game you only
can grow with a quadratic rate because the expansion speed is limited by
the unit speed. 

Timestamp n:  10 x 10
Timestamp 2n: 20 x 20
Timestamp 3n: 30 x 30
Timestamp 4n: 40 x 40

In the early game the growth is exponential because every settler is
able to find a place to build a new city.

Thomas
-- 
Thomas Strub  ***  eMail ue80@xxxxxxxxxxxxxxxxxxxxx
jb: people are stupid, they don't want to learn.