For Soren: suggestion for AI city trade algorithm

[Charlie]

My 2 euro cents on the matter. There would be a lot of factors I think of :[list=1][*]number of national citizens of the Civ[*]ratio of the Civ's natives vs. foreign citizens[*]history of Civ "owners" of city[*]great wonders[*]proximity to Capital/FP[*]buildings in city[*]proximity to other cities (effective city radius)[*]strategic location[*]access to resources (through road/harbor/airport)[*]improvements around the city (road/irrigation/mining)[*]resources in city radius (known at the time of trade)[*]maximum possible size/production/commerce[/list=1]
More comments on the numbered items :

• 1, 2 and 3 are most important. Any Civ is willing to get back its own people, isn't it ?
• 4 : makes sense if the wonder isn't obsolete
• 5 : of course !
• 6 : a city with banks, etc. is more valuable than a city without
• 7: I don't know much about this one. I certainly wouldn't like to get a city when there are others only 2 squares away.
• 8: this is a tough one (to code, at least). Think of Gibraltar...
• 9, 10 : do you want a big city without any road around that's going to revolt and starve because of overpopulation ?
• 11 : yes, I put "at the time of trade"
• 12 : who wants a city in the middle of tundra (except for resources) ?

Every criteria I listed above should have a cost value, depending on its importance. This is AFAIK what the Computer Player (CP -- I don't like AI, because of the "I" ) does for other trades, then it "converts" that value to a gold equivalent.

Other ideas ?

[player1]

Originally posted by Charlie
My 2 euro cents on the matter. There would be a lot of factors I think of :[list=1][*]number of national citizens of the Civ[*]ratio of the Civ's natives vs. foreign citizens[*]history of Civ "owners" of city[*]great wonders[*]proximity to Capital/FP[*]buildings in city[*]proximity to other cities (effective city radius)[*]strategic location[*]access to resources (through road/harbor/airport)[*]improvements around the city (road/irrigation/mining)[*]resources in city radius (known at the time of trade)[*]maximum possible size/production/commerce[/list=1]
More comments on the numbered items :

• 1, 2 and 3 are most important. Any Civ is willing to get back its own people, isn't it ?
• 4 : makes sense if the wonder isn't obsolete
• 5 : of course !
• 6 : a city with banks, etc. is more valuable than a city without
• 7: I don't know much about this one. I certainly wouldn't like to get a city when there are others only 2 squares away.
• 8: this is a tough one (to code, at least). Think of Gibraltar...
• 9, 10 : do you want a big city without any road around that's going to revolt and starve because of overpopulation ?
• 11 : yes, I put "at the time of trade"
• 12 : who wants a city in the middle of tundra (except for resources) ?

Every criteria I listed above should have a cost value, depending on its importance. This is AFAIK what the Computer Player (CP -- I don't like AI, because of the "I" ) does for other trades, then it "converts" that value to a gold equivalent.

Other ideas ?

Distance from Palace & Forbiden city should be driving force when doing city trade (it should be even more important then 1,2,3)

Why?

Because it varies form civ to civ.

I had so many games in which AI would be far more better is they tried to "optimize" their cities (swap them to get less corruption).



One more important thing:
It should always be on city for city base (no more then 25% difference in value), plus extra money for differnce.

Why?

So human won't exploit AI by bying or selling cities just for his money.


P.S.
I hope that Soren could get some nioce ideas form this.
There is no point of some option if it is not used by AI.
Right?

[Charlie]

I forgot :

• total population of city
• nearby cities' culture
• proximity of warring Civs and enemy military forces
• pollution in the city radius

I'm not sure about trading city for city only. Suppose : A and B are allied against C. C capture one of B's cities. A captures it back later. A should be able to sell back the city to B, without giving away one of its own. It could give the city back too, but it seems normal that A gets some payment for the city's liberation.

Exploiting the CP by selling worthless cities wouldn't be possible if worthless cities were, well, worthless A crappy city (1 pop, in the middle of wasteland, far from the palace/FP) isn't worth much if all parameters I tried to enumerate are taken into account.