AIs and the Art of War

[shortnmad]

My first run through on C2C at prince difficulty I've notices that the AI is dump as bricks and doesn't defend territory at all...I have a death ball with 2 heroes and elephants and the AI doesn't build any spear man but instead builds chariotrider and stacks then by 2s and 3s all over the place allowing me to farm XP without any damage being taken. Why the AI at all afk stacks 2-3 units about 20 in total and doesn't counter attack or defend properly like civ 4. Also the AI doesn't expand well as mentioned before, I have 11 city to the AI 4-5 and double their score.

Which version of C2C has a good AI to fight against without jacking up to deity and handicapping the crap out of myself?

 

[Arakhor]

I believe that turning off Terrain Damage in the BUG options should cut down on the healer spam, but I don't know what causes the AI to spam siege units like no tomorrow. There are so many siege units in my game, I'm sure it's slowing down turn processing.

 

[JosEPh_II]

My first run through on C2C at prince difficulty I've notices that the AI is dump as bricks and doesn't defend territory at all...I have a death ball with 2 heroes and elephants and the AI doesn't build any spear man but instead builds chariotrider and stacks then by 2s and 3s all over the place allowing me to farm XP without any damage being taken. Why the AI at all afk stacks 2-3 units about 20 in total and doesn't counter attack or defend properly like civ 4. Also the AI doesn't expand well as mentioned before, I have 11 city to the AI 4-5 and double their score.

Which version of C2C has a good AI to fight against without jacking up to deity and handicapping the crap out of myself?

Need more details than just running at prince difficulty level. The Options used Will make a huge difference in your game play. Also what version? The official last release v35? Or the current SVN version which has many "fixes" since the v35 release?

Without enough details say the AI is "dump as bricks" is basically meaningless. If you play vanilla BtS at Deity you should play Caveman2Cosmos at Deity or Nightmare Mode.

JosEPh

 

[shortnmad]

I did Hydromancerx recommendation on prince v35. I play prince+ normally.

Map: C2C_PerfectWorld2f (I like how it forms realistic biomes)
Size: (Use the biggest your computer can handle. Also use viewports to speed up the game on larger maps.)
Era: Prehistoric (You need to get the full effect from the beginning)
Speed: Snail (Anything faster is just too fast for this mod. I would do Eternity but its too slow for me).
New World Rules: Start in Old World (This makes it so I can found colonies later)
Pangaea Rules: (Either Choice is fine. I have yet to see an advantage to either other than more land wars on Pangaea)
Wrap Options: Cylinder (I just like this setting)
Team Rules: (I don't play on teams so it doesn't matter)

Options (Only listing the ones I check off)
- Start As Minor Civ (This is a MUST!! I love this option.)
- Usable Mountains (This is such an awesome mod. Not sure why its not just required.)
- Surround and Destroy (Great for city attacking).
- Advanced Diplomacy (It makes AI interactions so much more interesting)
- Barbarian Generals (Makes them more fun to battle)
- Assimilation (This is such a cool feature. It also is really needed to obtain non-native cultures)
- Great Commanders (Makes battles more fun)
- Culturally Linked Starts (I am not sure if this always works out. But it sounds good in theory)
- Advanced Economy (Not sure how it works, but it sounds cool).
- Realistic Culture Spread (I find this makes the game a bit more challenging. Which I like.)
- Religious Decay (A must have since there are TONS of types of religions.)
- Barbarians Always Raze (This is more of a personal preference)
- United Nations (A must have if you pick Mastery Victory)
- Advanced Espionage (Come on its spies! Why would you not want this?)
- Modern Corporations (Makes the most out of the Modern Era)
- Advanced Nukes (Shout out to Civ Fuehrer and his awesome mod!)
- Infinite XP (Personal preference)
- Developing Leaders (Its very cool to add-on new traits as you progress)

Victories (Only listing the ones I check off)
- Mastery (This is all you need since its all of them combined into one.)

 

[Dancing Hoskuld]

Start as Minors is known to handicap the AI because of the extra era C2C has.

 

[Arakhor]

And Advanced Economy will slow the game down even further, because of all the inflation calculations.

 

[Thunderbrd]

Need more details than just running at prince difficulty level. The Options used Will make a huge difference in your game play. Also what version? The official last release v35? Or the current SVN version which has many "fixes" since the v35 release?

Without enough details say the AI is "dump as bricks" is basically meaningless. If you play vanilla BtS at Deity you should play Caveman2Cosmos at Deity or Nightmare Mode.

JosEPh

Nah... he's actually right. There's a lot of problems that are simply not option specific at all. I CAN fix them all if I have infinite time handed to me at some point. In the interrim I'm thinking of what could be done for SOME quickfixes but there are issues that simply aren't quickfixes and won't be.

 

[JosEPh_II]

Nah... he's actually right. There's a lot of problems that are simply not option specific at all. I CAN fix them all if I have infinite time handed to me at some point. In the interrim I'm thinking of what could be done for SOME quickfixes but there are issues that simply aren't quickfixes and won't be.

Really TB?

@shortnmad,

As DH and Arakhor have stated several of your options will hinder the AI, but your biggest drawback is that v35 is now rather "old" vs the Updates provided thru the SVN.

And while Tb is "technically correct for v35" he's not totally correct. OIptions Do make a Big difference even for v35.

Mastery Victory setting also "dumbs the AI down". Too many victory goals for the AI to be effective. Fewer Victory settings leads to better AI performance.

One thing you didn't mention was number of AI at Game start. But if you selected the Old World map setting on the Largest PW2 available, then I would bet you put in at least 20 AI (probably more). And you have enabled Revolutions by not checking the No Revolution box. Rev sadly destroys the AI as the game progresses. Cool option for slowing down the Player but Bad "over all" for the AI to be effective opponents. But the Rev lovers will defend Rev even when the truth slaps them upside the head.

All that said next game try at least Emperor or higher difficulty level. Abandon Start as Minors and Rev. Try a PM map no larger than Huge and 12-15 AI. Forget about using Old World setting, use play everywhere instead. Colonies are meh. They don't suit C2C that well.

But do what you will as you're the one playing it after all. I just try to provide some insight into better game play from the Mod after almost 7 years of playing it from the start of v1 to current SVN version.

Good Hunting.

JosEPh

 

[Thunderbrd]

I just mean to say that there are severe strategic deficiencies in the AI, particularly in the areas he's talking about. No option can change those particular deficiencies.

Not that they might not overall perform better on particular options, just that many strategic problems don't have anything to do with any options. Such as their terrible judgement when it comes to what units to build. Ex: Their under-appreciation for spears even when faced with incoming death by mounted units.

Ultimately, the 'strategic' issues that he's pointing at have a lot to do with a failure of the AI to keep up with the proliferation of a variety of new and necessary unit roles to uphold and how to manage those unit roles and balance those needs appropriately. How to build and manage stacks properly.

Again... options have nothing to do with that. Sure, there are identified options like S&D that the AI doesn't know how to use to it's advantage. Sure, they can adjust how well they research but the problem I have is the same as what he's saying, it means you have to let them dwarf you on the civil side of the game (since they're good enough to do that when you set the game difficulty up and possibly even on noble) so that you can begin to have a challenge strategically in war because they're so stupid with their units that they need to be out-teching you to have a chance in the field.

It would be a much more fun game to play if they were good at keeping up with you civily and also good at challenging you in the field without having to be tremendously farther ahead technologically.


And again... I have plans to solve all of that but it's a massive project. And since that project is so integral to forward progress and so overwhelming, I don't see myself working up the will to sacrifice that many months of effort any time soon.

 

[shortnmad]

Thanks for the replies and suggestions.

 

[Arakhor]

And again... I have plans to solve all of that but it's a massive project. And since that project is so integral to forward progress and so overwhelming, I don't see myself working up the will to sacrifice that many months of effort any time soon.

Maybe it's a bit selfish, but perhaps a small change would be to lower the AI weight/lovefest that the AI have with siege weaponry. 40 siege towers and 70 trebuchets (I think it was) in one stack is just ridiculous.

 

[Thunderbrd]

Maybe it's a bit selfish, but perhaps a small change would be to lower the AI weight/lovefest that the AI have with siege weaponry. 40 siege towers and 70 trebuchets (I think it was) in one stack is just ridiculous.

Been considering this.

The problem is in the code and isn't likely solvable with AI weights. Most all AI weights are at 0 for units to begin with now and other factors are sometimes much more important, in this case in particular the percentage of a stack that the AI feels a city attack stack should be siege.

BUT there's a rudimentary dial I created in the globaldefines.xml that you can try to reduce yet further that would affect that. It's incapable of being a perfect fix because I believe many early siege in a stack results from rams being evaluated as being better than swords and other units for simply attacking a city (non-siege role) which may be resolvable roughly by increasing unit weights on sword units. My fear in using this particular dial is that it may create some unexpected mathematical results but someone can try it out to find the right balance.:

	<Define>
		<DefineName>C2C_ROUGH_BOMBARD_VALUE_MODIFIER</DefineName>
		<iDefineIntVal>0</iDefineIntVal>
	</Define>

Try taking it into the negative and see if it helps. Maybe -20 for a test run.

As for the spamming of healers, I think I may have asked the AI to overcompensate by making sure nearly all units in a large stack COULD be covered by a healer. Thing is, in retrospect, after a real battle many would be dead or others not even have been harmed. So really its overextending there. Again, a problem in the code.

However, what we could do to reduce the amount of healers needed would be to say, add 4 to all healing units' iNumHeal tags. You'd have to add the same amount across the whole board of them but it may be worth doing just to bring this under control without having to go into the code to fix the issue.

 

[Faustmouse]

However, what we could do to reduce the amount of healers needed would be to say, add 4 to all healing units' iNumHeal tags. You'd have to add the same amount across the whole board of them but it may be worth doing just to bring this under control without having to go into the code to fix the issue.

Wouldn't make it more sense to increase this by 300% instead of adding a plain +4? If an early healer with 2 iNumHeal gets +4 thats 200%, but if a late game high tech unit with iNumHeal 20 gets +4 that's only 20%.

 

[Arakhor]

Try taking it into the negative and see if it helps. Maybe -20 for a test run.

I'll give that a try, thanks.

 

[Arakhor]

So, I've made the suggested change and I've also adjusted the iNumHealSupport as Faustmouse suggested, e.g. 1 to 3, 2 to 6, 3 to 9 and so on. I'll have to see how that works in the game.

 

[Thunderbrd]

Wouldn't make it more sense to increase this by 300% instead of adding a plain +4? If an early healer with 2 iNumHeal gets +4 thats 200%, but if a late game high tech unit with iNumHeal 20 gets +4 that's only 20%.

I'm not 10000% sure that I'm stating the right tag here so without looking I just want to make sure we both understand that I'm talking about the tag that indicates how many units a given healer type can cover with healing support in a given round. The reason that in the early game you see so many healers is because the healers cover so few units at the beginning of the game. By simply adding +4 to all, you keep a reasonable gradual progression and don't change the shape of the chart of increase, just the Y axis upon which the chart is drawn so that all healer units cover 4 additional units. Already the amount of units later game healers can support (particularly with their additional amounts being added by more advanced promos) makes having only a few in a large stack perfectly acceptable.

I'm NOT indicating the strength of their healing ability - that's a different tag and shouldn't be messed with although I think the modmod got it absolutely right.

 

[Arakhor]

Would <iHeal> be the tag you mean? For example:

<HealUnitCombatType>
   <UnitCombatType>UNITCOMBAT_HEALS_AS_ANIMALS</UnitCombatType>
   <iHeal>5</iHeal>
</HealUnitCombatType>

 

[Thunderbrd]

Would <iHeal> be the tag you mean? For example:

<HealUnitCombatType>
   <UnitCombatType>UNITCOMBAT_HEALS_AS_ANIMALS</UnitCombatType>
   <iHeal>5</iHeal>
</HealUnitCombatType>

No. That's the healing amount not the amount of units that can be healed in a round. For whatever reason, my text editor is being a right now so I can't verify this but I THINK it's iNumHeal.

 

[Faustmouse]

I'm not 10000% sure that I'm stating the right tag here so without looking I just want to make sure we both understand that I'm talking about the tag that indicates how many units a given healer type can cover with healing support in a given round. The reason that in the early game you see so many healers is because the healers cover so few units at the beginning of the game. By simply adding +4 to all, you keep a reasonable gradual progression and don't change the shape of the chart of increase, just the Y axis upon which the chart is drawn so that all healer units cover 4 additional units. Already the amount of units later game healers can support (particularly with their additional amounts being added by more advanced promos) makes having only a few in a large stack perfectly acceptable.

I'm NOT indicating the strength of their healing ability - that's a different tag and shouldn't be messed with although I think the modmod got it absolutely right.

Yes we were thinking about the same tag then. I just wanted to make sure you keep in mind that +4 on a later unit is far worse than +4 on the earliest units.

 

[sunrisereader]

The AI seems to pick the repair promotion for siege engines instead of the ones that improve the ranged damage/city attack.