The key to AI greatness

[alva]

lol@ DeepO: knowing where you come from I didn't think you would like these changes, cultural diversity and all that.
But I agree with you, If all the AI are the same then certainly on some maps under the right conditions, they will be killer civs, I.E. unbeateble( maybe to strong a word ).
But most of the time I'm guessing they will all be little more then mediocre.
And this really is no fun at all.
When they're different from each other though, surely some will not make into the modern or even industrial age. But at least ther's a very good chance a few will go on the become superpowers.
So I would be quite reluctant to change much too, maybe a few tweaks here and there.
But what you can do though, is when you are going to play a pangea map for example, is change certain things that won't be as important (build naval springs to mind offcourse), afterall Firaxis made this list to work on all maps.

Sounds a bit like nature doen't it

[Arrian]

Thanks for the replies, Soren. Cool, I was right.

So the key categories to produce power, so far as I can tell are:

wealth
trade
production

happiness and science improvements are also good, but I think the priority needs to be on the three I listed.

Some testing may help. Perhaps we could make some tweaks to this and use it in a "Apolyton U" game.

-Arrian

p.s. Why hasn't Theseus, the seeker of "killer AI's" visited this thread yet?

[FrustratedPoet]

Originally posted by Arrian

p.s. Why hasn't Theseus, the seeker of "killer AI's" visited this thread yet?

I PM'd him about it in case he had overlooked the thread

... is he even online?

[DeepO]

Originally posted by alva848
lol@ DeepO: knowing where you come from I didn't think you would like these changes, cultural diversity and all that.

Yeah, sometimes you just have to have some conviction, and after that you try to find some 'proof' why you are right Indeed, I wouldn't like it if the Zulus start to build culture like crazy, let them overrun lesser civs in the ancient era, I'll gladly culturally flip their cities in the middle ages

But what you can do though, is when you are going to play a pangea map for example, is change certain things that won't be as important (build naval springs to mind offcourse), afterall Firaxis made this list to work on all maps.

Certainly... only choosing the right civs for some scenario would be good. And if you set the starting positions, you can generally predict which of the civs will do good in all games, and which won't reach the middle ages...

BTW, good to see you gained your avatar today, I'm hoping to lose my warlord status tonight

DeepO

[Theseus]

WHOA!!

Awesome thread, with Soren no less.

Let me digest...

alexman is da man!

[Theseus]

I'm going to need at least tonight to process 'build preferences' into my thinking...

I just want to say that the interactions of all features into rewarding gameplay balance is INCREDIBLY complex... my hat is off in another galaxy in respect to Soren and the Firaxis team.

[Txurce]

Alexman, thanks for a stimulating thread. As a Mac user with no editor to access, I found your chart in the original post to be invaluable. Now Arrian and others are running in the right direction with it.

This sems to be the best way to create as killer an AI as possible, and one that can be tested relatively easily and quickly. On Emperor, the AI's 20% production edge could theoretically make it competitive, if its focus was on shields to produce beakers. It might put the human in a position where war is (once again) the only way to the top.

Does anyone think that emphasis on offensive units would make an AI more dangerous? Despite the fact that this is flagged for six civs, I rarely see the sort of mounted forces that people on this forum routinely assemble.

[Lord Merciless]

Chinese, Persian, and Roman civ traits got changed in last second before the game was shipped. That should explain the discrepancies in the AI Civ's build priorities with their civ traits.

[Harovan]

Originally posted by Theseus
I'm going to need at least tonight to process 'build preferences' into my thinking...

But don't make it too hard, or else I'll have to start with Chieftain again, when I return to play .

[Arrian]

I was just thinking...

We all know the AI is incapable of using bombard units effectively. Therefore, they should not waste the shields building them. All civs should have "never build" for bombard units until such time as Firaxis provides a patch that changes how the AI uses them (if this ever occurs).

Whatcha think?

-Arrian

[DeepO]

They do use bombers... would putting no bombard mean no bombers as well? They are air units, sure, but would probably fall into both categories as well.

And, artillery gets used on defense, just stop too close with a ship to a city with art in it, and you'll see what happens.

DeepO

[Arrian]

Yes, they do use them defensively, but I've never seen more than 2 bombard units in any given AI city, and these will not bombard approaching stacks of troops. They will not be used offensively (well, ok, I did see the Greeks do it once, but it was a single catapult with a hoplite guard... ooooh, I'm scared now, Alex) and not even used properly for defense. The human knows that arty is only effective in numbers, so we put together stacks of artillery. These can be zapped around to pulverize invading troops. The AI will not do this.

So I'm saying maybe it's best to just tell the AI to forget it and build another spearman instead of a catapult.

-Arrian

[DeepO]

Arrian, I agree about the poor handling of artilleries by the AI, but they can use bombers relatively good. If these also fall under 'arty', I wouldn't put the build never option in.
Besides, the AI will only use 2 per city max, but will also build 2 per city max... and they are nice to capture

DeepO

[player1]

Even defensively, if AI would just learn to stack all it's arties in several most important cities (using ZERO-movment rails), it would be much better.

And of course, it things get to tough, to fall back with those arties to some other cities (so they won't get captured).

[phunny_pharmer]

Question for Soren-

Are all buildings with the 'wealth' flag rated equally by the AI. Will the AI choose ANY 'wealth' related building, or is it subject to some restrictions.

For instance, suppose it has the choice between an 80 shield building that produces 50% tax @ 2 upkeep, or a 120 shield, 50% tax 25% science @ 3 upkeep building. I think one of the mods had this scenario.

Now, which would a 'wealth' choosing AI build? Is there any way to *make* the AI build one more often then the other without using prereqs?