deity aw 4: the three musketeers

bobbyboy29 · Aug 3, 2011

obsolete said:
In any case I did some thinking on opponents. Two leaders I definitely want to see are Ragnar & Nappy. These are the two top fanatics when it comes to attack courage.

Can't say I'm quite in this league but from my few attempts at the other deity AW's I'd agree that attack courage should be a top priority. Civs with high attack priority are really just feeding you exp in this format, in my experience they tend to end up sending a constant stream of suicidal mini-stacks in waves rather than sit outside your city and build one giant SOD which will eventually overwhelm your defenses. The cost of replacing all of these units should severely slow them down as well. I think this would be pretty crucial.

Yamps · Aug 3, 2011

Check out these examples for more info on espionage: http://forums.civfanatics.com/showpost.php?p=10746250&postcount=33

You could easily get the discount down to 20% of the starting beaker cost. City Culture bonus is critical here, it's equal to 0.5*YourCityCulture/TotalCityCulture in the target city.

Possible plan:

Sitting Bull - Philosophical and Protective
Oracle CoL - Holy city and courthouse
Generate the great spy
Switch to Caste and get the great artist
Have a trap city prepared (optimally on a hill 3 tiles from your blocking city)
Spread your religion to the trap city and culture bomb it
Have spies stationed there for 5 turns
Let the most advanced AI capture it
Infiltrate the great spy
Prolonged profit

When you culture bomb the city, it's city culture increases by 4000. This move isn't absolutely necessary, the CC bonus will be at maximum value of 50% at the start without it. However, when the AI starts building the city the YourCityCulture/TotalCityCulture in the city will drop. Using the culture bomb, even if the AI gets to 500 culture, you'd get 0.5*4000/4500 = 44%. Of course, nothing from stopping you to build up some additional city culture the old fashioned way before they capture it.

It also isn't absolutely necessary to settle 3 tiles from your blocking city. This move is recommended though to fight for tiles near your trap city, you want it as small as possible.

DerangedDuck said:
....
If they built some culture in a city, but later lost control of that city, that city won't allow them to control any of the surrounding tiles. However, the culture that the city spilled onto the surrounding tiles might allow another of their cities to control those tiles.
....

Spread culture missions could help with that task too. If the TotalCityCulture is equal to say 4000, Culture Spread adds 5%, 200 culture. This mission also adds a certain amount of Plot Culture, somehow related to the size of the added City Culture. A small task force of ~5 spies can act as a small culture bomb to take over tiles. Note that this mission requires you to have some Plot Culture of your own on the target city's tile.

Example:

Yamps said:
Let's see how much for Feudalism. I'll assume:

-40% for different state religion and holy city -> 0.60
-50% for City Culture -> 0.50
+10% for distance -> 1.10
-50% for Stationary Spy -> 0.50
-20% for Espionage Point Spending -> 0.80

Feudalism has a cost of 1183 beakers, 1774 base EP cost. Modifiers: 100 -> 60 -> 30 -> 33 -> 16 -> 12. 12% of 1774 = 212 EP. That's 5.58 beakers for 1 EP, not bad at all. So, at this rate 3000 EP from that first spy would translate to 16740 beakers. Key thing to get that is to keep the City Culture bonus high and there are more ways to do it. IMO, 4000 culture from a great artist is the best way.

A point on Espionage Point Spending, 0.80 is a modest assumption. The multiplier is equal to (2x + y)/(x + 2y), x being their total espionage spending against all civs, y ours. Maximum discount is therefore 0.5, 2 is the worst case. In the early game, 3000 points from infiltration matters a lot. Let's say they only have 300 EP invested total and that we have 3000. The multiplier is then 0.57, 43% for 'Espionage Point Spending'. Let's see 1000 vs 3500. It's still nice, 31% for EPS.

EDIT: I'm guessing that spies would survive city capture. Not a critical part of the plan, but it deserves testing. They do get the -50% instantly if you gift the city in a normal game, I've done that.

GKey · Aug 3, 2011

The trap city must grow to size 2 before been captured. All in all not that easy task.
With some luck barbs can settle close enough. Not 3 tiles away but close. Still making most advanced AI getting it, sounds to me as kind of complicated task.

Yamps · Aug 3, 2011

GKey said:
The trap city must grow to size 2 before been captured. All in all not that easy task.
With some luck barbs can settle close enough. Not 3 tiles away but close. Still making most advanced AI getting it, sounds to me as kind of complicated task.

That's why I said on a hill.

Maybe if you use it as your main blocking city and then evacuate to the one 3 tiles behind.

Dirk1302 · Aug 3, 2011

It's kind of tricky. In my AW3 game i captured a barbcity and waited for some civ to divert to that city. It was the Zulu's, most advanced civ around. But sometimes that won't happen or it will take too long or some other civ decides they want it.

Blockcity isn't too easy to give up to the right civ as well. Most of the time their are units from multiple nations before it. Also Blockcity is usually placed in a way that it really blocks. If it falls you're trampled under foot. I don't think that'd be doable.

It just depends on the map, you can try to settle and give away more than 1 city. You don't have to hold it for that long, size 2 is ok. Border pop maybe too?

Yamps · Aug 3, 2011

Dirk1302 said:
Blockcity isn't too easy to give up to the right civ as well. Most of the time their are units from multiple nations before it.

It helps if you determine the order in which the AI moves. You can do that if you turn on the 'show enemy moves' option. Or if you choose your opponents, you can tell the order before you start the game.

It just depends on the map, you can try to settle and give away more than 1 city. You don't have to hold it for that long, size 2 is ok. Border pop maybe too?

If you culture bomb, you'll get the border pop.

Yes, as soon as the city hits size 2 it's good to go.

Snaaty · Aug 4, 2011

i like the idea of the "give away city". ideally it would not be near your block city, but 3 tiles away from your capital (=behind your block city) to reduce stealing costs for distance and to ensure the ai has absolutely no tiles to work

-> how to do it:

Spoiler :

Dirk1302 · Aug 4, 2011

^3 tiles from capital is awfully close. Early game you cannot easily dislodge all those AI units that'll be close to your capital. I think 6-10 tiles would be better so we can improve tiles near capital and inner core without being harassed. Bantu in AW3 is ideal. Stealing is a bit more expensive due to distance there but it's well outside te core of my empire. Even then the Zulu's shipped in a stack there once and i had to defend Awlil. You don't want that happening near capital before say ~500 AD.

vranasm · Aug 4, 2011

the city has to be in some distance to AI culture (or is it city centre?) for the AI to accept it anyway...

JammerUno · Aug 4, 2011

three words:

move the capital

Yamps · Aug 4, 2011

~10 tiles from the capital is quite okay, the distance penalty should be around 10%. 3 tiles from a blocking city guarantees that the trap city can get tile coverage from a blocking city when blocking city has 100 culture. If the trap city has 4000 culture, the first culture spread mission will then add 200 city culture to the trap city. The second mission 210. Run ~5 missions and the trap city won't have any tiles to work, missions will add plot culture too. I've done that with AI border cities in a WB test too, the culture mission adds 5% to the total city culture, not just your own. To enable it, you just need some plot culture on the tile.

EDIT: Hmm, I had more culture in the nearby city than 100, this way the trap city should still have the control of it's tiles on the other side. 500 culture in a blocking city would get another set of tiles. (to control the tile, one of your own cities needs to have the tile in it's cultural radius and you need plot culture dominance)

@vranasm

No problems with that in AW

obsolete · Aug 4, 2011

The trap city must grow to size 2 before been captured.

You guys keep forgetting that lots of leaders will still raze a city at pop-2.

I'm wondering if increasing the culture in that city only increases the odds of a decision-raze by the AI.

If we can have a code expert look through the AI's decisions on this matter...

r_rolo1 · Aug 4, 2011

Obsolete, the function you want to see is this one:

Code:

void CvPlayerAI::AI_conquerCity(CvCity* pCity)
{
	CvCity* pNearestCity;
	bool bRaze = false;
	int iRazeValue;
	int iI;

	if (canRaze(pCity))
	{
	    iRazeValue = 0;
		if (GC.getGameINLINE().getElapsedGameTurns() > 20)
		{
			if (getNumCities() > 4)
			{
				if (!(pCity->isHolyCity()) && !(pCity->hasActiveWorldWonder()))
				{
					if (pCity->getPreviousOwner() != BARBARIAN_PLAYER)
					{
						pNearestCity = GC.getMapINLINE().findCity(pCity->getX_INLINE(), pCity->getY_INLINE(), NO_PLAYER, getTeam(), true, false, NO_TEAM, NO_DIRECTION, pCity);

						if (pNearestCity == NULL)
						{
							if (pCity->getPreviousOwner() != NO_PLAYER)
							{
								if (GET_TEAM(GET_PLAYER(pCity->getPreviousOwner()).getTeam()).countNumCitiesByArea(pCity->area()) > 3)
								{
									iRazeValue += 30;
								}
							}
						}
						else
						{
							int iDistance = plotDistance(pCity->getX_INLINE(), pCity->getY_INLINE(), pNearestCity->getX_INLINE(), pNearestCity->getY_INLINE());
							if ( iDistance > 12)
							{
								iRazeValue += iDistance * 2;
							}
						}
					}


					if (pCity->area()->getCitiesPerPlayer(getID()) > 0)
					{
						if (AI_isFinancialTrouble())
						{
							iRazeValue += (70 - 15 * pCity->getPopulation());
						}
					}
                
			        if (getStateReligion() != NO_RELIGION)
					{
                        if (pCity->isHasReligion(getStateReligion()))
                        {
							if (GET_TEAM(getTeam()).hasShrine(getStateReligion()))
							{
	                        iRazeValue -= 50;                            
						}
							else
							{
								iRazeValue -= 10;
							}
						}
					}
                    
                    int iCloseness = pCity->AI_playerCloseness(getID());
                    if (iCloseness > 0)
                    {
                    	iRazeValue -= 25;
                    	iRazeValue -= iCloseness * 2;
					}
                    else
                    {
                    	iRazeValue += 60;
					}

					if (pCity->area()->getCitiesPerPlayer(getID()) > 0)
					{
						if (pCity->getPreviousOwner() != BARBARIAN_PLAYER)
						{
                            iRazeValue += GC.getLeaderHeadInfo(getPersonalityType()).getRazeCityProb();
						}
					}
					
					if (iRazeValue > 0)
					{
					    for (iI = 0; iI < GC.getNumTraitInfos(); iI++)
					    {
                            if (hasTrait((TraitTypes)iI))
                            {
                                iRazeValue *= (100 - (GC.getTraitInfo((TraitTypes)iI).getUpkeepModifier()));
                                iRazeValue /= 100;
							}
						}

                        if (GC.getGameINLINE().getSorenRandNum(100, "AI Raze City") < iRazeValue)
						{
							bRaze = true;
							pCity->doTask(TASK_RAZE);
						}
					}
				}
			}
		}
	}

	if (!bRaze)
	{
		CvEventReporter::getInstance().cityAcquiredAndKept(GC.getGameINLINE().getActivePlayer(), pCity);
	}
}

BTW most of the AI has a very low iRazeCityProb ( a lot of them has 0 ( that, btw, does not avoid razings ). GK has the highest with 75 , with Brennus, Nappy, Monty and Shaka with 50.

There are a lot of things that influence this besides the LH value though. I guess the ones that will influence you more is the closeness of the city to the AI and state of the AI finances ( they will raze more if they are in dire straits )

Fun trivia, but most likely useless for you: the AI will not raze cities before turn 21 ( independent of speed

) and will rarely raze one before it has 5 cities ( IIUC )

EDIT: The relation with distance ( directyl, without relation with closeness ) is that iRazeprob is added by the double of the distance if the distance is bigger than 12 ... and there is no direct relation with city size IIUC

the raze prob increases with upkeep ...

HuibertJan · Aug 4, 2011

obsolete said:
You guys keep forgetting that lots of leaders will still raze a city at pop-2.

I'm wondering if increasing the culture in that city only increases the odds of a decision-raze by the AI.

If we can have a code expert look through the AI's decisions on this matter...

An AI will never raze when

It has 4 or fewer cities
Fewer than 21 turns have elapsed since the beginning of the game
City is a holy city
City has an active worldwonder

An AI will raze when

The AI has a city near* and is in financial trouble and the city pop is fewer than 4.
The AI has a city near* and that city is farther than 9 tiles away and the city did not belong to barbarians.
The AI has no city near* and the target has more than 3 cities left and the city did not belong to barbarians.

At the end if the city did not belong to barbarians and the AI has a city near*
a random number compared is to the iRazeCityProb in the CIV4LeaderHeadInfos.xml

Note: near* probably means "same continent", did not verify.

HuibertJan · Aug 4, 2011

r_rolo1 said:
...snip

My code looks different. What sdk are you looking at and where can I find it?

My code is smaller, cleaner.
iRazeCityProb is used in a check on its own. The construction with iRazeValue is gone. No checks with shrines and religion.

r_rolo1 · Aug 4, 2011

I got it from my 3.19 install :confused:

I wonder if one of us has some not updated version of the code ...

But in the end your conclusions make sense regardless of the code ... atleast for the time frames the team is discussing ( and I really don't expect they will give a city with a wonder for the AI to pick

)

HuibertJan · Aug 4, 2011

HuibertJan said:
An AI will never raze when

It has 4 or fewer cities

Fewer than 21 turns have elapsed since the beginning of the game

City is a holy city

City has an active worldwonder

An AI will raze when

The AI has a city near* and is in financial trouble and the city pop is fewer than 4.

The AI has a city near* and that city is farther than 9 tiles away and the city did not belong to barbarians.

The AI has no city near* and the target has more than 3 cities left and the city did not belong to barbarians.

At the end if the city did not belong to barbarians and the AI has a city near*
a random number compared is to the iRazeCityProb in the CIV4LeaderHeadInfos.xml

Note: near* probably means "same continent", did not verify.

HuibertJan said:
My code looks different. What sdk are you looking at and where can I find it?

My code is smaller, cleaner.
iRazeCityProb is used in a check on its own. The construction with iRazeValue is gone. No checks with shrines and religion.

r_rolo1 said:
I got it from my 3.19 install I wonder if one of us has some not updated version of the code ...

But in the end your conclusions make sense regardless of the code ... atleast for the time frames the team is discussing ( and I really don't expect they will give a city with a wonder for the AI to pick )

I had accidently opened the code from the Civ4 vanilla install, instead of the BTS install.

The first half hasn't changed.

The second half has been transformed to a set of weights that either decrease or increase the probability of razing with the addition thatt he target city having state religion decreases the chance of razing.

obsolete · Aug 5, 2011

I'm also thinking, we should try to avoid enemy leaders who are aggressive or protective. Come to think of it, I also wouldn't want to fight vs any of the english leaders either.

Snaaty · Aug 5, 2011

on enemy leaders:

im heavily biased towards random leaders...

Dirk1302 · Aug 5, 2011

Yes, it's more fun that way.

deity aw 4: the three musketeers

I was saying boo-urns...

Deity

Emperor

Deity

Deity

Deity

Deity

Deity

Deity

King

Deity

Deity

King of myself

Chieftain

Chieftain

King of myself

Chieftain

Deity

Deity

Deity

Similar threads