Better economic AI (V. 0.5)

[Turinturambar]

Yeah, I fixed that problem in the new version (basically finished, I just need the proper time to upload it) by forcing civs to actively spread their religion if they have only one holy city and aren't currently following that religion. Works quite nicely for epic Ashen Veil/Order showdowns.

Esus has the problem that in base FFh2 the AI is only allowed to use the spread esus spell if it already follows that religion. You can see why it rarely gets adopted. Fixed in the new version by allowing the AI to spread it if it either follows esus or has the holy city.

 

[[to_xp]Gekko]

great to see this is still being developed. furthermore, the new stuff should end up being also in Wild Mana quickly, and that will eventually be in base FFH. so... thanx for making FFH a hell of a lot better

 

[Neomega]

Hell yeah, I'm excited to see this AI mixed with wild mana's magic AI in base FFH. (and then maybe on to FF)

 

[Opera]

I'm eager to see your new version, Turin

 

[lemonjelly]

So am I. May I use this in Fallen Ages?

x

 

[bread smith]

For what version is this?
I'm looking for an AI tweek for FFH2 040y or 040z.

 

[Turinturambar]

I finally got around to uploading the new version. It is linked in the first post. Sorry for taking so long.

@bread smith: it should work with older versions as long as they are not too old.

@lemonjelly: Sure use whatever you want from it.

 

[Secret_Squirrel]

Got a download straight away from me, need to see what I can "borrow".

 

[Tholal]

AI will no longer build forts or build workshops before the guilds tech.

Why the restriction on building forts? How does having the Guilds tech make forts a more viable option?

 

[[to_xp]Gekko]

awesome! you should ask a moderator to update the name of the thread

a couple questions:

1) did you notice if the AI handles ships better thanx to it understanding the crew promotions?

2) will Kuriotates build settlements? the fact that they are expansive makes settlers a good deal for them, although of course they should not overdo it ( having few cities means that what you produce in each one is very important )

 

[Turinturambar]

Why the restriction on building forts? How does having the Guilds tech make forts a more viable option?

It was meant only for workshops. The AI will never build forts with this mod, as they can't use them properly and they are pretty pointless anyway.
Actually they don't use workshops anymore at all now too(unless fallow), since even after guilds, a workshop is rarely a good improvement and the AI perfoms better without them.

Gecko;8449736]awesome! you should ask a moderator to update the name of the thread

a couple questions:

1) did you notice if the AI handles ships better thanx to it understanding the crew promotions?

2) will Kuriotates build settlements? the fact that they are expansive makes settlers a good deal for them, although of course they should not overdo it ( having few cities means that what you produce in each one is very important )

I asked a moderator to change the name the last time and never got a reply.

Regarding 1:
I usually test on Pangeae, so ships are a minor concern. But the AI is better at settling islands, since it will use the cargoenhancing promotion. It probably still is very bad at doing intercontinental invasions or playing archipelago style maps.

Regarding 2: No they won't build settlements at all. Even though Cardith is expansive a settlement costs the AI: The cost of the settler, 2 military units as garrison, worker turns for improving the useless land near the settlement, increased maintenance in the core cities and possibly reducing the amount of workable tiles in the core cities.

The benefit of a settlement is the aquiring of usually 1 new happy/health resource.

That is usually a very bad deal. Check the screenshot, the Kurios built their five cities, then teched rapidly due to high happycaps and steamrolled the Clan of Embers with bronze centaur chargers. They are usually now at the top of the score ranks in my testgames, whereas before they usualy got easily killed by a neighbour.

 

[Secret_Squirrel]

Looking through the code seeing what I can "merge" I have found a minor mistake:

Line 1790: if pPlayer.hasBonus(gc.getInfoTypeForString('BONUS_MITHIRL')):

Edit: CvGameUtils.py

 

[Turinturambar]

Thanks.
So the AI will not take into account whether it has access to Mithril or not, when deciding which military building it should build first in a city. Oh well, shouldn't be such a big deal .

 

[Secret_Squirrel]

Yeah, definitely not game breaking. I especially like your training building code, its similar to something I was looking to do but you just made the job a whole lot easier!

 

[Senethro]

I always enjoy your updates, will give some feedback soon. Is your AI supposed to be played with Aggressive AI turn on or off?

Can you guess whether 0.7 will merge well with Wild Mana?

 

[[to_xp]Gekko]

It was meant only for workshops. The AI will never build forts with this mod, as they can't use them properly and they are pretty pointless anyway.
Actually they don't use workshops anymore at all now too(unless fallow), since even after guilds, a workshop is rarely a good improvement and the AI perfoms better without them.



I asked a moderator to change the name the last time and never got a reply.

Regarding 1:
I usually test on Pangeae, so ships are a minor concern. But the AI is better at settling islands, since it will use the cargoenhancing promotion. It probably still is very bad at doing intercontinental invasions or playing archipelago style maps.

Regarding 2: No they won't build settlements at all. Even though Cardith is expansive a settlement costs the AI: The cost of the settler, 2 military units as garrison, worker turns for improving the useless land near the settlement, increased maintenance in the core cities and possibly reducing the amount of workable tiles in the core cities.

The benefit of a settlement is the aquiring of usually 1 new happy/health resource.

That is usually a very bad deal. Check the screenshot, the Kurios built their five cities, then teched rapidly due to high happycaps and steamrolled the Clan of Embers with bronze centaur chargers. They are usually now at the top of the score ranks in my testgames, whereas before they usualy got easily killed by a neighbour.

I disagree about workshops, I think that after you have smelting and guilds they are wicked good. 3H1C is hardly crappy, comparable to a aristograrian farm and way better than a cottage at that time of the game ( since it would take loooong for the cottage to grow )

great news about ships

regarding workers wasting time building improvements in settlements, I'm pretty sure that in Wild Mana they don't ( probably came from Skyre's mod actually ) , so you might want to take a look at that if it's any use to you.

 

[Turinturambar]

Yeah I would play with aggressive AI on, Acheron disabled, a nonislands map and whatever else you like.

It's trivially easy to merge with Wildmana, but I fear that my AI is not optimized for his DLL changes. I ran a testgame with my gameutils file and his latest beta and my AI went into strike due to hording too many units and had trouble sending out explorers. They wouldn't even go for the huts right next to their culture zone of control.

So I'm confident that Sephi will pick the parts he likes and which fit his Dll changes. Just as I did for the update, the trainingyard building code is based off Sephi's template for example.

 

[Secret_Squirrel]

Unless I have misunderstood the code, you have limited many alternatives to axemen and champions to 2 despite (imo) them having a valid role on the battlefield and providing much needed variety in the troops fielded in those eras.

Just out of curiosity, do people think its worthwhile sacrificing flavour (in the stack composition of AI) just because the standard option has the potential to be slightly better?

 

[Turinturambar]

Unless I have misunderstood the code, you have limited many alternatives to axemen and champions to 2 despite (imo) them having a valid role on the battlefield and providing much needed variety in the troops fielded in those eras.

Just out of curiosity, do people think its worthwhile sacrificing flavour (in the stack composition of AI) just because the standard option has the potential to be slightly better?

Which one would you say have a valid role on the battlefield and should be fielded in significant numbers?

All the units I blocked either cost a lot more than their axeman alternative (90 vs 60) or are significantly worse than champions (can't use weapons etc).
As long as the AI still builds a few of them, flavor is preserved. Also note that I limit them to 3 not 2.

 

[Secret_Squirrel]

If i remember rightly, the Crusader and Paramander are both only 10 more than a Champion, which is minimal when considering you would need to build 12 or so Champions to have enough left over for a "Free Champion".

They start with Medic, which is useful and leads to an earlier option to upgrade to March when they reach 2XP. They can use weapons of all kinds and above all afaik they are Disciple Units, meaning that against units with a bonus vs Melee (Shock,
Champions, etc) they have quite an advantage compared to the standard Champion. Not to mention the use of Disciple Unit killing promotions, a far less widespread.

Some of the limits I agree with, such as the Monk as they arent particularly great units to build a large number of.