Hi! For a long time I had this questions in mind about the AI policies ("never build" and "build often") coming down to: Can we find a set of policies that lets the AI make saner choices?? After a lot of testing in the last months, I make a proposal here for something that might actually improve the AI opponents or at least makes them appear more rational.
On top of that, additional ideas sneaked in. Some of them are connected to the initial goal, and others that try to modernize the Civ3 experience for single player. Therefore, I plan this effort to be a three-part-exercise:
A scenario based on the original ruleset that only offers the new AI policies and nothing else;
An expanded version of the AI Policies mod with some additional changes and some added content solely related to AI policies;
A scenario with the full set of ideas that includes AI policies as well as some changes to tech effects, naval unit stats, resources, a few new buildings, without changing the basic game experience.
I'd like to start with items 1. and 2.
Complementary AI Build policies
What I call AI policies are the settings for "build never" and "build often" in the Conquests Editor. They are all over the place. There are a lot of misconceptions floating around. The effective number should be a maximum of six, as stated by one of the original developers. I tested what buildings are effected by which, but I omit the details here.
The best way to make use of them I found was to give the AI some guidance in addition to what the AI does without any of those flags set. They prefer cheap of costly. They take trait flags into account. There is some randomness involved. They need lots of happiness to prevent entertainers. Some improvements are always built regardless of the policy settings, e.g. barracks, courthouses.
So the lead questions became:
For a scientific civ, a "build science often" flag doesn't make sense because they choose libraries and universities even without any flags more often than non-scientific civs. Religious loves culture buildings, commercial loves trade buildings etc.
So I came up with a pattern of complementary policies that selectively emphasizes improvements civs tend to neglect or help the civs' traits.
There is a universal push for workers and growth. AI makes mostly bad use of workers, but with more workers they improve more tiles which provides an overall benefit. Growth feeds into this by enabling larger cities. And for larger cities, the AI needs happiness if they don't already prioritize culture or trade buildings. The Growth flag stays deactivated only for agricultural civs because they already prefer growth.
Offensive units are flagged for aggressive civs because the end up at war more often but then lack the attack power to hurt their enemies.
Commercial civs have the wealth flag checked for banks, stock exchanges, but wealth can also trigger marketplaces, and in some cases the happiness flag is deactivated, e.g. India: Because India is religious, they build Temples and Cathedrals more often, they get the Wealth flag because they are commercial, therefore they build marketplaces more often and don't need to spend even more shields on happiness buildings.
These kinds of arguments have been applied to all civs.
Additional Features for AI Improvement
The main problems with the AI mentioned again and gain are as follows:
The Theatre is a strategic alternative to the temple for the human player and also a replacement targeted to the AI for the overly expensive colosseum which has been removed and converted to a small wonder.
Public Gardens provide additional happiness and may not be of huge interest for the human player but the AI will happily choose to build it and gain additional happiness that they miss out on because of their perpetual 0% luxury spending.
To mitigate the infamous "Courthouse in capital" problem, Courthouses provide reduction of war weariness and Police Stations protect from propaganda. The human player may still avoid Courthouse in the capital but the AI at least get some additional happiness from it in certain circumstances.
The hardest problem was treatment of the wonder addiction. I eventually settled on additional prerequisites like resources and buildings to guide the AI into building less or at least more useful wonders.
This comes with the cost of making it harder for human players to build wonders. This is not the perfect scenario to achieve a cultural victory. I tried to mitigate this caveat by adding +2c to most great wonders, and +1c for most small wonders.
I also introduced the Burial Site (Ceremonial Burial) as a temple alternative. It only produces 1c for 40s.
The graphics for all new buildings (Burial Site, Theatre, Public Gardens) have been extracted from the original "Conquests", so the visual style is very consistent.
All changes should be reflected in the Civilopedia. Please let me know if you wish more detailed documentation.
The ZIP contains two BIQ scenario files and a folder which need to be placed in the Scenario folder in the Conquests section of your Civ3 installation.
On top of that, additional ideas sneaked in. Some of them are connected to the initial goal, and others that try to modernize the Civ3 experience for single player. Therefore, I plan this effort to be a three-part-exercise:
A scenario based on the original ruleset that only offers the new AI policies and nothing else;
An expanded version of the AI Policies mod with some additional changes and some added content solely related to AI policies;
A scenario with the full set of ideas that includes AI policies as well as some changes to tech effects, naval unit stats, resources, a few new buildings, without changing the basic game experience.
I'd like to start with items 1. and 2.
Complementary AI Build policies
What I call AI policies are the settings for "build never" and "build often" in the Conquests Editor. They are all over the place. There are a lot of misconceptions floating around. The effective number should be a maximum of six, as stated by one of the original developers. I tested what buildings are effected by which, but I omit the details here.
The best way to make use of them I found was to give the AI some guidance in addition to what the AI does without any of those flags set. They prefer cheap of costly. They take trait flags into account. There is some randomness involved. They need lots of happiness to prevent entertainers. Some improvements are always built regardless of the policy settings, e.g. barracks, courthouses.
So the lead questions became:
- What does a civ prefers to build anyway?
- What suits the civs traits most?
- Where does this civ get happiness from?
For a scientific civ, a "build science often" flag doesn't make sense because they choose libraries and universities even without any flags more often than non-scientific civs. Religious loves culture buildings, commercial loves trade buildings etc.
So I came up with a pattern of complementary policies that selectively emphasizes improvements civs tend to neglect or help the civs' traits.
There is a universal push for workers and growth. AI makes mostly bad use of workers, but with more workers they improve more tiles which provides an overall benefit. Growth feeds into this by enabling larger cities. And for larger cities, the AI needs happiness if they don't already prioritize culture or trade buildings. The Growth flag stays deactivated only for agricultural civs because they already prefer growth.
Offensive units are flagged for aggressive civs because the end up at war more often but then lack the attack power to hurt their enemies.
Commercial civs have the wealth flag checked for banks, stock exchanges, but wealth can also trigger marketplaces, and in some cases the happiness flag is deactivated, e.g. India: Because India is religious, they build Temples and Cathedrals more often, they get the Wealth flag because they are commercial, therefore they build marketplaces more often and don't need to spend even more shields on happiness buildings.
These kinds of arguments have been applied to all civs.
Additional Features for AI Improvement
The main problems with the AI mentioned again and gain are as follows:
- Ignorance of the happiness slider
- Useless spending of shields
- A severe case of wonder addiction
The Theatre is a strategic alternative to the temple for the human player and also a replacement targeted to the AI for the overly expensive colosseum which has been removed and converted to a small wonder.
Public Gardens provide additional happiness and may not be of huge interest for the human player but the AI will happily choose to build it and gain additional happiness that they miss out on because of their perpetual 0% luxury spending.
To mitigate the infamous "Courthouse in capital" problem, Courthouses provide reduction of war weariness and Police Stations protect from propaganda. The human player may still avoid Courthouse in the capital but the AI at least get some additional happiness from it in certain circumstances.
The hardest problem was treatment of the wonder addiction. I eventually settled on additional prerequisites like resources and buildings to guide the AI into building less or at least more useful wonders.
This comes with the cost of making it harder for human players to build wonders. This is not the perfect scenario to achieve a cultural victory. I tried to mitigate this caveat by adding +2c to most great wonders, and +1c for most small wonders.
I also introduced the Burial Site (Ceremonial Burial) as a temple alternative. It only produces 1c for 40s.
The graphics for all new buildings (Burial Site, Theatre, Public Gardens) have been extracted from the original "Conquests", so the visual style is very consistent.
All changes should be reflected in the Civilopedia. Please let me know if you wish more detailed documentation.
The ZIP contains two BIQ scenario files and a folder which need to be placed in the Scenario folder in the Conquests section of your Civ3 installation.
