Game engines. Civ3 vs Chess

[wohmongarinf00l]

though i am an amateur in algorithm design i think both chess and civ3 require dynamic algorithms to be truly good. a chess AI can do this and make sacrifices because the search space for the dynamic algorithm is smaller and gets smaller as there is only one desired result.
civ3 ai is a different animal. a truly dynamic algorithm would require tons of resources in computational power and memory.
each civ could have a certain eng goal and it would be able to evaluate decisions and try to predict the outcome that best brings it closer to the outcome in X number of turns. since the path changes depending on what happens every turn, the path would have to be continually recalculated.
to minimise chances of players "learning" what the endgoal of each civ is, it could be factored in the random variable so each civ has equal chances of getting a certain "task" (spaceship, UN, conquest, domination, high score)....where there are lots of civs, two civs could have the same task.
the trick is, how to hide the task from the human player.

as it currently is, i suspect it uses greedy algorithms for everything. it tries to win something everyturn and assumes that by accumulating all these wins, it will win in the end. otherwise it would have the concept of ranking importance of cities (both its own and opponents). its battle strategies are always based on taking the most weakly defended cities and when u have a massive stack of offensive units u hardly get it attacked. that's why u can walk from one end of a civ to the other with that one huge stack without meeting an ai initiated counterattack on the stack. of course when u attack u get defenders.....

 

[alver]

Originally posted by MuddyOne
I think CIvs AI could be much more effective if each civ was given one additional trait- method to try and win by, and strategies where developed to try and accomplish it. Like the advice given to people who can win at cheiftain but stall at more difficult games. Pick a goal and go for that goal.

That's a great idea! I guess the AI strategies are not as diverse now as your suggestion would demand, but in some ways, I think it looks like a few of the civs are aiming for different victory conditions.

Take the persians; in my games they always have large armies and often get powerful, expanding their lands against other AI civs. They are going for domination, perhaps.

The babylonians tend to get a lot of culture points. They seem to go for the cultural victory.

These are the best examples I can think of at the moment. I like these distinctions, and would love to see them made more deliberate, as you suggest.

 

[wohmongarinf00l]

Originally posted by alver

Take the persians; in my games they always have large armies and often get powerful, expanding their lands against other AI civs. They are going for domination, perhaps.
The babylonians tend to get a lot of culture points. They seem to go for the cultural victory.

this is what i mean by players learning the endgoal of a particular civ. it makes the whole thing pointless.

 

[alver]

Well.... knowing the Persians' style (for example) doesn't always make it possible for me to beat them, but I see your meaning. I can always count on the Persians not to get in the way of my cultural victory, and that makes it a little easier for me.

 

[MuddyOne]

Alver, they have different production schedules currently. It would require a lot of work on the AI though to make it in to stategies.

Wohmongarinf00l, good point. Instead of a fixed victory condition to stive for, a random one, that is weighted for the civilization's strengths and weaknesses would be my preference. This way the likelyhood of a militaristic civilization peacefully building through the game- loosing all benefits from being militaristic- becomes less likely.

 

[Zouave]

The point in fact is this:

I have never heard anything like the complaints about the AI when Civ 2 came out. Civ 2 gave the appearance of a smarter AI; the Civ 3 AI not only cheats it does various thoroughly dumb things that result from sloppy programming and an unwillingness to fix them, not to mention a lack of historical knowledge by the programmers.

Examples have been posted far too many times to do so again here.

The Civ 3 AI is inadequate and disappointing compared to Civ 2 - not compared to chess. We all expected better after over five years of alleged development between the games.

 

[alver]

I know that's your opinion; you've stated it 932 times already...