Killtech
Discutator
ok i'll use this thread for all my observations concerning the AI and will provide my analysis here. so let's get started...
i've started a new game recently with the newest version of AND. marathon speed, giant map, 40 civs, start as minors and increasing difficulty option starting at prince. bad choice for me but good enough to compare AI with me (normally i play at deity). i ended the game at 1500 BC after completing third world wonder and entering world builder to see that i was much to powerful to be defeated. however i get the feeling the AI has slightly improved but the core problem remains: AI cripples its economy with too much military. a direct compare: i have 4 cities and 5 military units; the AI mostly had less cities and more than 10 units; my cities had most food & production buildings completed (about 10+ buildings); the best AI city had 6 buildings. maybe it's because the minor civ start (all civs are at war, no diplomacy at until writing) have to check this next time. i need more observation to discuss this further.
another issue i've observed: AI, barbarians and the player. this just doesn't work as it is rendering the game too easy for experienced players. the reason is following strategy: due to the barbs are aggressive and will attack without any check on combat odds you can exploit this by moving units onto forest near them => you win and farm xp. it's very important in early game sages because small cites don't provide much of a defence bonus. furthermore stationing your units outside your borders on defensive positions gives you more time to ready your defences and concentrate your military in case of a enemy stack. (and yes, my cities mostly have no garrison - especially early on). all in all barbs are a good training for my units (and i'm only worried if they are missing). on the other hand i see frequently the AI losing new cities to black stacks.
obviously the option 'raging barbarians' is quite game breaking because of this huge imbalance. it usually kills several civs early on while it leaves me with a highly trained military and no relevant damage. i recommend reworking the barbs. they should be unable to build up any stacks and be less aggressive but instead focus on pillaging to force players into the offensive. the barb units should behave a bit like animals... every one roaming independently. but not attacking enemies unless they are injured. only if multiple barbs are near a city by chance they can try to attack it.
---------------------
here come some new observations on the AI along with some suggestions:
0. Advanced start
AI makes very very(!) bad choices here. if you find some time you could rework that maybe, Afforess. priories are following:
0. set the capital
1. one defensive unit (it's necessary for the AI but i mostly don't spent points on that)
2. place a single city or a settler if no good place for 2nd settlement is available.
3. increase population in capital by one.. sometimes in 2nd settlement too (huge economy boost for the start)
4. tech depending on resources visible
5. one worker can't hurt (or an work boat if fish is in city radius)
6. culture improvements (if the city will have low hammers and a valuable resource (food) is out of culture borders)
7. tech to speed up the discovery of a religion
1. Early Expansion
first of all it expands too much. settlers are take very long to complete, stop growth and the new cities they fond cost maintenance. and at the top of that you need to build more military to ensure their safety. one should not expand greater than his economy can handle. pretty early in the game the when the AI still had only one the capital and it was still size 1 it started to build settlers! i think a prerequisite should be at least a size 2 city to consider first expansion (in the case the capital is the only city). and as long as a city doesn't reach size 3 or 4 building further settlers shouldn't be considered.
2. Later Expansion
for later games the AI should first check how many 'developed' and 'core' cities it has compared to 'small towns' to decide over further expansion. as a rule the amount of small towns & conquered cities in the empire should never go over 60% of all cities (but lower is better).
3. Military Buildings
namely 'walls', 'bandit's outpost' and 'barracks'. the first two increase city defense the last gives xp. the AI gave these buildings too high priority. they were build mostly before any production building and short after the grainy. the walls may be okay but only if raging barbarians are active. the others were definitively too early.
4. Ships
the AI build a galley pretty early which didn't make much sense to me. though it was just a single one.
5. City Placement
one fonds new settlements if one finds a good place for it. the AI builds cities because it has some settlers left. even if it's in the desert and no matter how the economy and revolution status look.
6. The Forge
AI build 'wheelwright' and 'stoneworkers hut' quite fast- which is good. but it don't want to build any forges although it certainly should. okay it's quite busy with all the revolutions but still should build it. it didn't complete any commerce buildings either (no bazaar, jewelery, baskermasters shop, lighthouse, ...). this seems to be the point at which the AI stops to play good.
7. Fish Traps without Fish
in the current version fish traps have no effect if you lack any sea-resource. the AI has not even one in its cultural borders but yet it build one. it's nothing big but it should calculate in the bonus depending on the available resources.
as it seems the is expanding far beyond its economic potential causing severe revolutions everywhere. furthermore the new cities cost quite some maintenance but give little commerce back since they are very undeveloped. additionally more military is require for protection of these settlements. form a certain point on thee AI has no capacities to spare for further city improvements anymore and is unable to keep up with the player.
i've started a new game recently with the newest version of AND. marathon speed, giant map, 40 civs, start as minors and increasing difficulty option starting at prince. bad choice for me but good enough to compare AI with me (normally i play at deity). i ended the game at 1500 BC after completing third world wonder and entering world builder to see that i was much to powerful to be defeated. however i get the feeling the AI has slightly improved but the core problem remains: AI cripples its economy with too much military. a direct compare: i have 4 cities and 5 military units; the AI mostly had less cities and more than 10 units; my cities had most food & production buildings completed (about 10+ buildings); the best AI city had 6 buildings. maybe it's because the minor civ start (all civs are at war, no diplomacy at until writing) have to check this next time. i need more observation to discuss this further.
another issue i've observed: AI, barbarians and the player. this just doesn't work as it is rendering the game too easy for experienced players. the reason is following strategy: due to the barbs are aggressive and will attack without any check on combat odds you can exploit this by moving units onto forest near them => you win and farm xp. it's very important in early game sages because small cites don't provide much of a defence bonus. furthermore stationing your units outside your borders on defensive positions gives you more time to ready your defences and concentrate your military in case of a enemy stack. (and yes, my cities mostly have no garrison - especially early on). all in all barbs are a good training for my units (and i'm only worried if they are missing). on the other hand i see frequently the AI losing new cities to black stacks.
obviously the option 'raging barbarians' is quite game breaking because of this huge imbalance. it usually kills several civs early on while it leaves me with a highly trained military and no relevant damage. i recommend reworking the barbs. they should be unable to build up any stacks and be less aggressive but instead focus on pillaging to force players into the offensive. the barb units should behave a bit like animals... every one roaming independently. but not attacking enemies unless they are injured. only if multiple barbs are near a city by chance they can try to attack it.
---------------------
here come some new observations on the AI along with some suggestions:
0. Advanced start
AI makes very very(!) bad choices here. if you find some time you could rework that maybe, Afforess. priories are following:
0. set the capital
1. one defensive unit (it's necessary for the AI but i mostly don't spent points on that)
2. place a single city or a settler if no good place for 2nd settlement is available.
3. increase population in capital by one.. sometimes in 2nd settlement too (huge economy boost for the start)
4. tech depending on resources visible
5. one worker can't hurt (or an work boat if fish is in city radius)
6. culture improvements (if the city will have low hammers and a valuable resource (food) is out of culture borders)
7. tech to speed up the discovery of a religion
1. Early Expansion
first of all it expands too much. settlers are take very long to complete, stop growth and the new cities they fond cost maintenance. and at the top of that you need to build more military to ensure their safety. one should not expand greater than his economy can handle. pretty early in the game the when the AI still had only one the capital and it was still size 1 it started to build settlers! i think a prerequisite should be at least a size 2 city to consider first expansion (in the case the capital is the only city). and as long as a city doesn't reach size 3 or 4 building further settlers shouldn't be considered.
2. Later Expansion
for later games the AI should first check how many 'developed' and 'core' cities it has compared to 'small towns' to decide over further expansion. as a rule the amount of small towns & conquered cities in the empire should never go over 60% of all cities (but lower is better).
3. Military Buildings
namely 'walls', 'bandit's outpost' and 'barracks'. the first two increase city defense the last gives xp. the AI gave these buildings too high priority. they were build mostly before any production building and short after the grainy. the walls may be okay but only if raging barbarians are active. the others were definitively too early.
4. Ships
the AI build a galley pretty early which didn't make much sense to me. though it was just a single one.
5. City Placement
one fonds new settlements if one finds a good place for it. the AI builds cities because it has some settlers left. even if it's in the desert and no matter how the economy and revolution status look.
6. The Forge
AI build 'wheelwright' and 'stoneworkers hut' quite fast- which is good. but it don't want to build any forges although it certainly should. okay it's quite busy with all the revolutions but still should build it. it didn't complete any commerce buildings either (no bazaar, jewelery, baskermasters shop, lighthouse, ...). this seems to be the point at which the AI stops to play good.
7. Fish Traps without Fish
in the current version fish traps have no effect if you lack any sea-resource. the AI has not even one in its cultural borders but yet it build one. it's nothing big but it should calculate in the bonus depending on the available resources.
as it seems the is expanding far beyond its economic potential causing severe revolutions everywhere. furthermore the new cities cost quite some maintenance but give little commerce back since they are very undeveloped. additionally more military is require for protection of these settlements. form a certain point on thee AI has no capacities to spare for further city improvements anymore and is unable to keep up with the player.
By wild I mean barbarian units which spawned in the fog of war and were not built in a city. I am going to play around with their default and other AI actions to see what happens.
