[freeciv-ai] Re: Measure the AI ability to harvest resources

["Per I. Mathisen" <per@xxxxxxxxxxx>]

On Sun, 24 Nov 2002, Raimar Falke wrote:
> I'm currently updating my unit agents patch. To measure the
> performance I'm using the attached patch. It measures the resources
> which a player collected with all his cities. Leading "2:" is from
> freelog. Than turn and than playername. Next 3 numbersare the food,
> shield and trade collected to far. Next 3 luxury, tax and science. s
> is the food*FOOD_WEIGHTING + shield*SHIELD_WEIGHTING +
> trade*TRADE_WEIGHTING.

Interesting. I hope you ran those AIs on 'hard', tho ;)

But we all knew the AIs settler code was stupid beyond the call of bad
coding. I've pretty much fixed the problems that we had with the AI
wasting production on stuff other than settlers (notably coinage).
Although it still _seems_ the AI produces stupid things, that is just
filler while it waits for enough food to be able to produce more settlers.
And since it switches production without penalty this isn't a problem.

There are other problems, though:
  - The AI doesn't optimize cities for food. Raimar, please update your
CM-in-server patch. It will fix this, and until it is fixed, CMA assures
that a good comparison between settler agent and server/settlers.c isn't
really possible.
  - The AI doesn't find the right spots, and travels too far before making
a city, often skipping a good spot it will settle later.
  - The AI is way too fond of making boats and going overseas to make
cities, wasting a lot of time.
  - The AI stupidly assumes both harbour and offshore when placing cities.
  - The AI reserves a too big field around its cities, making cities at a
too big distance to each other.

I have an old patch which rewrites the city building code completely which
should fix all but the first problem. I had placed it on hold because I
first required fixes to roles, which we've now done. I'll see if I can
update it once I get some time.

  - Per