Re: [Freeciv-Dev] Superficial analisis of AI research speed.

[Tony Stuckey <stuckey@xxxxxxxxxxxxxxxxx>]

On Sat, Apr 03, 1999 at 12:22:35AM +0000, Manuel Gutierrez Algaba wrote:
> Today, I've played at another nice game of freeciv, following these
> premises:
> - it's better to have lots of cities than to have few cities
> - it doesn't matter how well equiped your cities are 

        This is pretty axiomatic.  With lots of cities, you get more land
worked in a shorter period of time.  It's not until late in the game that 
Library/University, etc and Marketplace/Bank etc grow to be a significant
productivity enhancer.
        Also, with small cities, you don't have to worry about unhappysize
so much.

> So I chose despotism , and for a long time I run under it. 
> The surprising thing happened when I did the switch to Republic 
> ( as I always do ), then science rate rose up to 60 % ( sixty per cent ).

        The AI will switch to Republic quite early.  This can be a problem
in some games, as that makes exploration with military units cause a lot of
unhappiness.

> Most surprisingly , I developed science as fast or even faster 
> than my full grown counterpart ( a nice AI ), and my empire was
> composed of a lot of small ( 5,6,4...) cities. After a while (10-20
> -30 turns , I can't figure out now )
> science rate fell down to normal (?) 17 % .

        The science cost depends on how many technologies you have
researched already.  Thus you would expect it to fall if your empire is not
expanding research speed as fast as that cost is expanding.

> Then, I have these conclusions:
> - AI speed of research  relies in its way of deploying as many 
> cities as possible without necessarily  equipping them .

        Yes.  Where human players often choose the opposite.  They make
roads for more trade, build up cities to about the Aqueduct size, etc.
        The proper strategy will depend on map sparseness.  A large map
with few players will draw even harder toawrds the "no cities above size 6"
style play.

> Even so, it's quite interesting and fine to have found the key
> of that impressive AI. I have the save-games of it. But It'd be
> interesting to experience that strategy with more people,... or
> to play at it a bit more.

        Sigh.  Now the AI has to get even better. :)
-- 
Anthony J. Stuckey                              stuckey@xxxxxxxxxxxxxxxxx
"When I was young, the sky was filled with stars.
 I watched them burn out one by one."  -Warren Zevon