C3C flavors can make the AI smarter!

[alexman]

At last, we can now make Religious AI civs beeline for Monarchy, Scientific AI civs grab Literature ASAP, Seafaring civs build curraghs (maybe) and other trait-specific science and build decisions come true!


The new C3C feature of 'flavors' in the editor can associate each civilization, building, and technology with a flavor (up to 7 flavors). So if we create a flavor for Religious civs, and another flavor for Scientific civs, and another flavor for Seafaring civs, et cetera, we can get different behavior for each trait!!!

Can you say smarter AI for the AU mod?

We humans have developed some civilization-specific strategies. Besides what I mentioned above, would you care to brainstorm with me about ways to put these 7 flavors to good use in the epic game? None of them are currently used.

[donZappo]

Can you use these flavors to specify civilizations to focus on building specific classes of wonders (agricultural, industrious attributes...) or just specific buildings? I'm surprised that these 'flavors' weren't allowed before C3C or even included in the core epic game as it truly helps civs to be distinctive!

-donZappo

[alexman]

You have to specify each wonder with its own flavor (can't do classes of Wonders), but you can give all Agricultural Civs and all Agricultural Wonders the 'Agricultural' flavor.

Each flavor has a percentage association with each other flavor. So Flavor 1 can have a 50% association with Flavor 2, for example. That means that Wonder with Flavor 1 is half as likely to get built by a civilization of Flavor 2.

Before moving forward with this idea, we would have to test and see what happens if a technology or building has two flavors. Do the percentages multiply, or does one of them get chosen? So if a building has both flavor 2 and flavor 3, (and say flavor 2 and 3 both have a 50% association with flavor 1), is a civilization in flavor 1 50% or 25% less likely to build that building?

I'll be away from civ for a whole week, but maybe some kind volunteer can do some tests in the meantime?

[DrSpike]

If it is in the editor does it imply that the developers have used/will be using the feature?

[Fosse]

I won't have Conquests until after the Holidays, but I have to say that this idea, alexman, is incredibly exciting to me.

I just want to make sure that I understand correctly... each Civ can have one flavor, so a decison would have to be made for each Civ as to which of its traits gets flavored. Correct?

Can flavors have negative modifiers? Can we make certain buildings less desirable as well as more to certain nations?

Also, I haven't followed the AU discussions much lately (in part becaue my PtW CD broke) so I know little about the new panels and upcoming games, but I see a nice set of flavor courses as being lots of fun.

Brilliant thinking, Alexman.

[alexman]

Each civ can actually have one or more flavors, and the same for techs and buildings.

Flavors can be only negative modifiers, but I'm sure there are ways of getting around this, if only we knew how multiple flavors affect the AI.

Again, some testing (or clarification from Breakaway) is needed as to what the AI does when multiple flavors are involved. (I can't test this before a week from now)

Flavors are already used in the Fall of Rome and Medieval Scenarios, but in those cases each civ, building, and tech, has at most one flavor.

[Fosse]

Originally posted by alexman
Flavors can be only negative modifiers

Did you mean positive?

[alexman]

No, I mean 'negative', as in they can't make the AI want to research something more than if it didn't have any flavors - only less or equal.

Say flavor 1 has a P percent relationship to flavor 2. P can be anything between 0 and 100.
If P = 0, a civ with flavor 2 will never reseach a tech in flavor 1.
If P = 100, a civ with flavor 2 will research a tech in flavor 1 with the same priority as if the tech had no flavor at all.
If P = 50, a civ with flavor 2 us half as likely to research a tech in flavor 1 than if the tech had no flavor at all.

[ZargonX]

Hmm... this definately holds a lot of potential. I'll try and take some time this week to start tinkering with them, but with the holiday coming up, it might take a little longer

[Nor Me]

Re: C3C flavors can make the AI smarter!

Originally posted by alexman
At last, we can now make Religious AI civs beeline for Monarchy, Scientific AI civs grab Literature ASAP

You could do that before, using unbuildable UUs.

If this is really a negative effect then balancing it to get a positive one is going to be tricky.

I doubt flavours work in the way they say so testing obviously going to be a good idea and it might not even be practical.

[alexman]

Yes, your unbuildable UUs was a great idea.

However, it didn't have much flexibility in the amount of value you can add to techs, and it didn't affect buildings, only techs.

[Catt]

I wondered how long it would take alexman to start a discussion on how flavors might enhance the AU Mod. Now awaiting word on how effective the new editor tool is at enhancing AI gameplay.

I suppose, alexman, that you're canceling your week away to instead explore the joys and possibilities of "flavors?" And Nor Me has, presumably, now acquired a copy of C3C?

Catt

[Nor Me]

No. Pre-ordering it is not enough.

[Fosse]

Originally posted by alexman
No, I mean 'negative', as in they can't make the AI want to research something more than if it didn't have any flavors - only less or equal.

Say flavor 1 has a P percent relationship to flavor 2. P can be anything between 0 and 100.
If P = 0, a civ with flavor 2 will never reseach a tech in flavor 1.
If P = 100, a civ with flavor 2 will research a tech in flavor 1 with the same priority as if the tech had no flavor at all.
If P = 50, a civ with flavor 2 us half as likely to research a tech in flavor 1 than if the tech had no flavor at all.

Got it. Thank you for the clarification.

Nor Me is right about the balancing being more tricky, with negative modifications only. But, hey... I have a lot of faith in you guys.

Must get Conquests....

[Theseus]

Wow, this could be very cool. Ultimately, we could HUGELY increase the chances of AI civs (given the right environment) developing into KAIs.