[freeciv-ai] Re: (PR#2644) AI needs 5400 years to get Republic

[Thomas Strub <ue80@xxxxxxxxxxxxxxxxxxxxx>]

On Sun, Jan 26, 2003 at 07:17:47AM -0800, ue80@xxxxxxxxxxxxxxxxxxxxx via RT 
wrote:
> On Mon, Jan 06, 2003 at 12:07:33PM -0800, Raimar Falke via RT wrote:
> > On Mon, Dec 30, 2002 at 10:35:02AM -0800, ue80@xxxxxxxxxxxxxxxxxxxxx via RT 
> > wrote:
> > > On Mon, Dec 30, 2002 at 10:10:56AM -0800, Per I. Mathisen via RT wrote:
> > > > 
> > > > [ue80@xxxxxxxxxxxxxxxxxxxxx - Tue Dec 24 15:35:14 2002]:
> > I just want to remind you that this problem was also observed in the
> > first contest. With a high research cost the AI didn't research any
> > tech which would enable it to build good units and so it built
> > baracks. This also wasted all trade -> 100% tax and no science.
> 
> Ok, i think its the fallback where AI builds only barracks and settlers
> when it doesn't know what to build. Think we should change it to another
> fallback.

And here the new fallback:

  if (can_build_unit(pcity, best_role_unit(pcity, F_TRADE_ROUTE))) {
    pcity->currently_building = best_role_unit(pcity, F_TRADE_ROUTE);
    pcity->is_building_unit = TRUE;
  } else{
    pcity->currently_building = B_CAPITAL;
    pcity->is_building_unit = FALSE;
  }

I tried 120 games with that:
42 times rep before -2000
44 times rep -1500 - -1950
23 times rep -1000 - -1450
10 times rep -50 - -950
1 times Republic 1 AD

Problem is that AI isn't expanding that fast, but think this should be
done outside the fallback. 
 
Thomas
-- 
Thomas Strub  ***  eMail ue80@xxxxxxxxxxxxxxxxxxxxx
Wenn Du nicht programmieren kannst und Dir für Arbeit zu schade bist:
Werde Berater, Analyst oder organisiere Kongresse.