Diplomacy by numbers - spreadsheet

Bibor

Doomsday Machine
Joined
Jun 6, 2004
Messages
3,124
Location
Zagreb, Croatia
XML Diplomacy and leader flavors for Civilization 5


jpbar81 did a terrific job, so I'm linking his spreadsheet:

JPBAR81's AI LEADERS SPREADSHEET DOWNLOAD

***

For all of you that don't have the expansion yet, here are the latest values for all leaders (all DLCs included).

Here's a google spreadsheet link:

https://docs.google.com/spreadsheet/ccc?key=0AsTxFrznhhOFdGZvOVRqXzVJWkRaUlYtaV9HQjB6eFE



List of changes from vanilla:

Spoiler :
Changelog:

Alexander
MajorCIV WAR from 7 to 6

Askia
Deceptive from 4 to 3

Augustus
Deceptive from 7 to 6
Build Recon from 4 to 3
Build units - Mobile from 5 to 4

Bismarck
MajorCIV WAR from 7 to 6

Elisabeth
MajorCIV WAR from 5 to 4

Gandhi
MajorCIV WAR from 3 to 2

Harun
Deceptive from 7 to 6
CS ignore from 6 to 3

Hiawatha
Deceptive from 7 to 5

Oda
Deceptive from 7 to 6

Ramkhamhaeng
MajorCIV WAR from 6 to 5
Deceptive from 7 to 6

Suleiman
MajorCIV WAR from 6 to 5

Washington
MajorCIV WAR from 5 to 4
Deceptive from 7 to 5

Isabella
MajorCIV WAR from 7 to 6
Deceptive from 6 to 5

Pachacuti
MajorCIV WAR from 7 to 5
Deceptive from 7 to 6

Kamehameha
MajorCIV WAR from 6 to 4

Harald Bluetooth
Deceptive from 5 to 4



How to read the table

The listed AI values represent how likely an AI leader is to:
- decide upon a way to win
- react to what you and the other AIs do
- react to wars, wonders being built, city-states etc.
- build units, buildings, wonders and what type

The higher the value, the more likely the AI will act according to that strategy. For example, Alexander has a very high score in pursuing a diplomatic victory. He might not take that route, but its very likely. Oda Nobunaga has a very low "Warmonger hate" score. He is very unlikely to hate you if you're a warmonger.

There's still a die roll involved (and other factors like distance from your civ etc.) but these tables should give you a very solid base on which you can plan your game so it affects diplomacy in the way you want it to.

Exact columns mean the follwing:

Competetiveness

Victory: chance of getting the "they think we try to win in a similar fashion" diplomatic penalty
Wonder: chance of getting the "they covet wonders we built" diplomatic penalty
Minor civ: chance of getting the "we compete for same city-states'" diplomatic penalty
Boldness: unknown
Diplobalance: unknown, possibly chance for getting diplomatic penalties for attacking or denouncing a nation you have declarations of friendship with.


Diplomacy

Warmongerhate: chance of diplomatic penalties for wiping out AIs, city-states and starting wars
Denouncewillingness: chance of denouncement, based also on number of active penalties
DoFwillingness: chance of wanting declarations of friendship, based also on active bonuses
Loyalty: chance of them backstabbing someone (war or denouncement)
Neediness: how likely they are go ask for resources/gold with DoF active
Forgiveness: chance of the AI forgiving old grudges (diplomatic penalties)
Chattiness: how likely they open up the screen and comment.
Meanness: unknown


Flavor

Offense/defense/city-defense: how likely they are to invest into offensive/defensive units or city defenses (walls, castles, military bases).
Military training: chance of building Barracks, Armories, Stables etc.
Recon/ranged/mobile/naval/navalrecon/air: Chance to build certain unit types
Navalgrowth/navaltileimprovement/waterconnection Chance to build lighthouses, seaports, fishing boats, harbors
Expansion, Growth, Tile improvement, Infrastructure: how likely they are to build settlers, workers, make tile improvements, roads
Production, science, gold, culture chance of building production, science, gold, culture buildings and focus on working those tiles
Happiness chance of building happiness buildings
GP, Wonders: first unkown, could be chance of building gardens or employing specialists, the other one is for chance of building wonders
Religion: unused (leftover from CIV4), currently affecting chance of going piety branch (positive) and going space victory (negative)
Diplomacy, Spaceship: how likely they are to go for UN or Science victory
Nuke: how likely they are to build Manhattan project and to build&use nukes.


Major civ approach

Neutral/friendly/guarded/hostile/afraid/war: How likely they are to decide upon having one of the stated attitudes, considering all other modifiers already in place
Deceptive: how likely their stated attitude does not reflect their real attitude. This is exactly the same model as was used in CIV4 (can declare at friendly, pleased, cautious) but otherwise does not affect current diplomacy (an AI that can declare war at Friendly and is thus deceptive, will still give you 300 gold for a luxury resource).
CS approach ignore/friendly/protective/conquest: how likely the AIs not to care / invest into (and do quests) / defend (civ X declares protection over CS Y) / conquer a city state.


How to read the table - Example leader: Ramesses

He's not likely to declare war/denounce you if you're close to victory.
He's likely to hate you for building lots of wonders.
He's likely to hate you (even denounce) if you declare wars often and wipe out civs (City-states count as a full civs)
He'll likely to offer and/or accept Declarations of Friendship.
Whatever reason he had to hate you in the past, he's not likely to ever forgive you.
He won't build many units. Of those that he does, most will be archers, artillery, mounted units and defensive units like pikemen or anti-aircraft guns.
He's going to have well-developed lands tile-improvement-wise.
His cities will contain workshops, granaries, watermills, markets, banks, culture buildings, as well as colisseums, circuses and theatres.
He will build as many wonders as he can.
He's most likely to go for a spaceship victory.
He will use nukes if he can.
Diplomatic relations are likely to deteriorate quickly with him (for whatever reason) but he is unlikely to declare war even then.

How to deal with Ramesses:
He's difficult to please and it is very likely that whatever you do he'll hate you eventually. Military-wise he's a pushover and you can take him out with sword/gunpowder units accompanied with siege. You don't need to fear his naval or air force, because its not likely to be ever up and running. Be wary of declaring on him after he completes manhattan project and has uranium, because he'll nuke you no problem. He's probably one of the best civs to completely wipe out or puppet up because his lands and cities are going to be filled with buildings, wonders and proper improvements.
 

Attachments

  • leader-noexp.png
    leader-noexp.png
    42.3 KB · Views: 82,576
That's genius, making Ghandi almost twice as likely to build nukes as ANY OTHER CIV.

How historically accurate and immersive.
 
I like how, on a scale of 1 - 10, Ghandi's love for nukes goes up to 12. Seriously, its the only number on there above 10 (and there are only two tens for that matter).

Great job making this table!
 
I like how, on a scale of 1 - 10, Ghandi's love for nukes goes up to 12.
Maybe the nuke flavor is a reverse scale? With the 2pt flavor modifier, Ghandi rolls a 10-14 for nukes, but a 10 or higher could completely disable the flavor in this case.

An extreme value like this suggests some sort of method for bypassing the underlying mechanics.

What I found interesting was Wu Zetian. Few things are as immersion crushing as when you come across a Wu Zetian who rolled a 2 for expansion and tile improvement each, and a 8 for wonders, only to have (maybe) one wonder built, no additional cities and no tile improvements by the Renaissance or later.
 
They must have changed Alexander. I remember his conquest of CS used to be one of the highest values but now he is ranked very high on protective. I also have noticed him going after diplomatic victory a lot after the patch.
 
I am going to assume that you realise that the Ghandi referred to in that Wikipedia article is Indira Ghandi, female Prime Minister of India, who, in fact, was not related to Mahatma Ghandi.

If we based a civilization's likelihood to build nukes on whether the leader historically would have built nukes, America would never have them.
 
Great work Bibor!

For the diplomacy stats that does explain some of the oddball behaviors; if an AI rolled very high on WarmongerHate for example I imagine they would be very prone to hating you for helping defend them in a war.

Some of the CS numbers probably need to be adjusted to make AI civs compete for CS favor more aggressively. It seems like only a few of the AIs in my games make a concerted effort to compete for CSs. With the vanilla game as it stands, there is only one good strategy for CSs: "compete for them!" and more AIs should be pursuing that good strategy than less ignoring them, which is just counterproductive.
 
12 for nukes means either

1. its outside the allowed range so therefore never build nukes

2. a developer set it to 12 as a joke and then forgot to set it back and it got live by mistake

FWIW, I've never seen Ghandi nuke anyone in any of my games.
 
What I find interesting in this graph are the victory numbers. It just seems strange to me that only a few AIs are competitive and will try to win, while the rest will just be content to sit there.

I don't know if this is new to civ (obviously I'm a very new civ player, look at my post count) but this just seems like a strange decision. Shouldn't every AI be competitive and try to win the game?
 
12 for nukes means either

1. its outside the allowed range so therefore never build nukes

2. a developer set it to 12 as a joke and then forgot to set it back and it got live by mistake

I also think it's a joke but was deliberatly left in: peaceful and being bashed around until he gets nukes.

As for the nuking itself, remember he has to have Manhattan as well as access to uranium. I've seen Gandhi go on a conquest spree (usually successful) in some of my games. I'm not sure he used nukes though (he probably would've if he had uranium).
 
Gandhi is crazy for nukes. In almost every game with India they have been the second civilization (after me) to finish the Manhattan Project.
 
Hm. What do you think the best AIs are, based on this chart?

It looks like Askia, Alexander, Oda Nobunaga, Suleiman, and Montezuma are toughest, based on the fact that they have 8+ offense, so they will likely build units that the tactical AI is not complete fail with.

Of these, Montezuma and Oda Nobunaga are most deceptive. Montezuma also only has a 2 for wonder-building, so he is unlikely to waste production. So if it weren't for companion cavalry and hoplites being scary, I would say Montezuma. However, Montezuma has an 8 for city state conquest, so he will likely waste all his time losing troops to city states.

So Greece wins, because Greece:
1. Classical UUs
2. High Offense (Hoplite/Sword Production instead of archer/catapult)
3. 3 Ranged (Lack of ranged units is bonus for AI effectiveness)
4. High Mobility (coupled with CCs = dangerous)
5. High Meanness (lol)
6. 4 at growth (lower than average!) which means more hammers early for more units
7. 7 at War/Hostile (likely to attack w/ military advantage)
8. 4 at conquest for city states (unlikely to waste troops to fighting CS)


Downsides:
7 at wonderbuilding (likely to waste hammers on this)
 
Top Bottom