Tweet
One of these days (maybe Christmas break?) I want to dive deeply into the code for the AI diplomacy. For me, the new diplomacy system is still the most intiguing development of Civ IV and I'm always looking for new ways to navigate the shifting sands. To that end, here is some quick work I did on the AIs' favorite civics choices. Really, it's just a compilation of the basics stats, but I think it may reveal some interesting strategies.
If I'm repeating someone else's work, please let me know and guide me in its direction.
The Premise
AI diplomacy is the effect caused by any number of "factors" on the AIs attitude towards another player. Factors are those reasons given for the AIs attitude towards another player when you mouse over the AI leaderhead. (e. g. "You declared war on us!") Each of these also feature a "weight," the plus or minus number after the factor which denotes the degree of the factor's effect on the attitude.
Without looking at the code, we can make some assumptions about the factors and weights from reasoning. 1) There are a finite number of factors. That is, the AI cannot invent new reasons to like or dislike you, it's not that smart--yet. Most of these factors the average player will be familiar with, but there are still some surprises. 2) The triggers of the factors are rigid. That is for example, a distaste for your choice of religion is never random, it is caused when the AI finds that your state religion is not the same as its own. The surprises, and the need for deeper inspection, come with 3), the weights are determined largely by AI "personality," where a weight of '0' is non-appearance of the factor.
(The fourth big effect of diplomatic attitude is "memory," which is the personality's tendency to retain an applied weight. For example, Montezuma seems never to forget that you declared war on him, while Gandhi seems never to care that you stopped trading with him.)
Favorite Civics
One of the tricks used to affect personality in fake historical figures and cause conflict between players (AI-AI, AI-human, and very possibly human-AI) is by giving them "favorite civics." In my observation, these are one of the more complicated factors and the factor that can most easily sneak up on an unsuspecting player.* But one that can also be used to great advantage if the player is aware of it.
Most simply, each AI leader has at least one favorite civic. All other players using that civic get a positive diplomatic factor; all players using an alternative get a negative diplomatic factor. (Remember, weights of these factors are largely determined by personality--in the case of favorite civics, I presume entirely so.)
Note that the factor does not seem to be triggered unless it is declined for an alternative. If the civic is not yet available to the player, the AI will not penalize that player for failing to adopt it.
The Types
Like with traits, which are utilized by the human assuming the role as the leader, favorite civics, which are not, divide the range of leaders into categories.
Again, in the main game each leader has one favorite civic and triggers the diplomatic factor if the player chooses an alternative. Thus, if the AI's favorite civic is Slavery, it will not have a factor concerning a player's adoption of Universal Suffrage. The former is in the labor category, the later in government category and not alternatives to one another. Therefore, the leaders must be separated according to two categories: first, their "bias," second their "favorite civic."
(Note that this is a v2.08 Warlords categorization!!!)
Government-biased
Monarchists
Authoritarians (Police State)
Aristocrats (Representation)
Democrats (Universal Suffrage)
Legal-biased
Bureaucrats
Nationalists
Feudalists
Libertarians (Free Speech)
labor-biased (an underpopulated category)
Elitists (Caste System)
economy-biased
Mercantilists
Free Marketeers
Communists (State Property)
religion-biased
Clericalists (Organized Religion)
Theocrats
Ecumenists (Free Religion)
The Interpretations
The largest group, as before Warlords, is the Monarchists, though their membership's been drastically reduced. The smallest groups, with one member, are the Libertarians (Washington) and the Elitists (Wang Kon).
Certainly, these groups can be seen as natual allies. For example, Cyrus and Mehmed II are both likely to stick with Vassalage longer and will mutually reinforce their preference for one another with positive diplomatic factors. Other groups are less likely. Although Brennus may share Organized Religion with either of the Egyptian leaders, he might not share the same religion, nullifying the positive effect of the favorite civic.
If all other variables are equal (including the impossibilities of equal numbers of players and equal time), some interesting phenomena pop up. The most obvious is that the Monarchists have the best diplomatic advantage, with a 4:9 chance that they will gain favor from other AI players. Washington and Wang Kon suffer the most with zero chance of gaining favor from other AIs. Democrats fair poorly, too, in Civ IV with a 1:9 chance of getting diplomatic favor from other AIs.
But consider when you, the (presumably) human player adopts a civic. If you adopt Monarchy, you have a 5:9 chance of getting a positive diplomatic affect, if you use Organized Religion, it is 1:2 (3:4 if both Egyptians are playing as AI). But if you adopt Caste System--the civic that hamstrings Wang Kon--you have a 1:0 chance of pleasing someone else. In other words, Caste System is the only 'safe' civic. One more reason it's one of my new personal favorites.
Keep in mind that you can influence your allies to change civics. With this, much like religion, you have more control over international politics.
Happy hunting!
*I speak from experience!
If I'm repeating someone else's work, please let me know and guide me in its direction.
The Premise
AI diplomacy is the effect caused by any number of "factors" on the AIs attitude towards another player. Factors are those reasons given for the AIs attitude towards another player when you mouse over the AI leaderhead. (e. g. "You declared war on us!") Each of these also feature a "weight," the plus or minus number after the factor which denotes the degree of the factor's effect on the attitude.
Without looking at the code, we can make some assumptions about the factors and weights from reasoning. 1) There are a finite number of factors. That is, the AI cannot invent new reasons to like or dislike you, it's not that smart--yet. Most of these factors the average player will be familiar with, but there are still some surprises. 2) The triggers of the factors are rigid. That is for example, a distaste for your choice of religion is never random, it is caused when the AI finds that your state religion is not the same as its own. The surprises, and the need for deeper inspection, come with 3), the weights are determined largely by AI "personality," where a weight of '0' is non-appearance of the factor.
(The fourth big effect of diplomatic attitude is "memory," which is the personality's tendency to retain an applied weight. For example, Montezuma seems never to forget that you declared war on him, while Gandhi seems never to care that you stopped trading with him.)
Favorite Civics
One of the tricks used to affect personality in fake historical figures and cause conflict between players (AI-AI, AI-human, and very possibly human-AI) is by giving them "favorite civics." In my observation, these are one of the more complicated factors and the factor that can most easily sneak up on an unsuspecting player.* But one that can also be used to great advantage if the player is aware of it.
Most simply, each AI leader has at least one favorite civic. All other players using that civic get a positive diplomatic factor; all players using an alternative get a negative diplomatic factor. (Remember, weights of these factors are largely determined by personality--in the case of favorite civics, I presume entirely so.)
Note that the factor does not seem to be triggered unless it is declined for an alternative. If the civic is not yet available to the player, the AI will not penalize that player for failing to adopt it.
The Types
Like with traits, which are utilized by the human assuming the role as the leader, favorite civics, which are not, divide the range of leaders into categories.
Again, in the main game each leader has one favorite civic and triggers the diplomatic factor if the player chooses an alternative. Thus, if the AI's favorite civic is Slavery, it will not have a factor concerning a player's adoption of Universal Suffrage. The former is in the labor category, the later in government category and not alternatives to one another. Therefore, the leaders must be separated according to two categories: first, their "bias," second their "favorite civic."
(Note that this is a v2.08 Warlords categorization!!!)
Government-biased
Monarchists
- Alexander
- Louix XIV
- Ragnar
- Catherine
- Huayna Capac
Authoritarians (Police State)
- Montezuma
- Shaka
- Genghis Khan
Aristocrats (Representation)
- Augustus Caesar
- Julius Caesar
- Napoleon
- Victoria
Democrats (Universal Suffrage)
- Frederick
- Gandhi
Legal-biased
Bureaucrats
- Kublai Khan
- Peter
- Qin Shi Huang
Nationalists
- Bismarck
- Churchill
Feudalists
- Cyrus
- Mehmed II
Libertarians (Free Speech)
- Washington
labor-biased (an underpopulated category)
Elitists (Caste System)
- Wang Kon
economy-biased
Mercantilists
- Roosevelt
- Tokugawa
Free Marketeers
- Mansa Musa
- Hannibal
Communists (State Property)
- Mao Zedong
- Stalin
religion-biased
Clericalists (Organized Religion)
- Brennus
- Hatshepsut
- Ramses II
Theocrats
- Isabella (formerly an Authoritarian)
- Saladin
Ecumenists (Free Religion)
- Asoka
- Elizabeth
The Interpretations
The largest group, as before Warlords, is the Monarchists, though their membership's been drastically reduced. The smallest groups, with one member, are the Libertarians (Washington) and the Elitists (Wang Kon).
Certainly, these groups can be seen as natual allies. For example, Cyrus and Mehmed II are both likely to stick with Vassalage longer and will mutually reinforce their preference for one another with positive diplomatic factors. Other groups are less likely. Although Brennus may share Organized Religion with either of the Egyptian leaders, he might not share the same religion, nullifying the positive effect of the favorite civic.
If all other variables are equal (including the impossibilities of equal numbers of players and equal time), some interesting phenomena pop up. The most obvious is that the Monarchists have the best diplomatic advantage, with a 4:9 chance that they will gain favor from other AI players. Washington and Wang Kon suffer the most with zero chance of gaining favor from other AIs. Democrats fair poorly, too, in Civ IV with a 1:9 chance of getting diplomatic favor from other AIs.
But consider when you, the (presumably) human player adopts a civic. If you adopt Monarchy, you have a 5:9 chance of getting a positive diplomatic affect, if you use Organized Religion, it is 1:2 (3:4 if both Egyptians are playing as AI). But if you adopt Caste System--the civic that hamstrings Wang Kon--you have a 1:0 chance of pleasing someone else. In other words, Caste System is the only 'safe' civic. One more reason it's one of my new personal favorites.
Keep in mind that you can influence your allies to change civics. With this, much like religion, you have more control over international politics.
Happy hunting!
*I speak from experience!