Good news is since 1.5, you can easily play a tall game where you have 5-10 planets, and conquer the galaxy. No need to REX and micro every planet.
Another close look at how the AI builds up its worlds: 1.6 edition
- Thread starter GC13
- Start date
-
We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
Good news is since 1.5, you can easily play a tall game where you have 5-10 planets, and conquer the galaxy. No need to REX and micro every planet.
Ah not that much as fanatic authoritarian and with the expansion tradition you only need to relocate 3 pops per colony.
A small price considering the "outstanding" ability of the AI to handle my planets.
Unfortunately the thing I was referencing wasn't a Mod.
For Stellaris it would just use the has_building trigger, I think it is, with a small weight, probably like 0.05. I'm not sure how the ai_will_build part of the actual building entry works, it might be used to determine the weight of the building overall, or it might just be a 'if over 1, AI might build and the actual deciding logic is hidden in the engine' sort of thing.
If you set your sector to force respect of tile resources and ban it from redevelopment you should be okay; at that point there's little that the AI can do to mess things up. The problem comes when the unchained AI is allowed to build up its own worlds all by itself, when it really shows that it has a defective decision making process.
I think sectors are flawed a bit. Sectors should combine multiple planets into single entity (like adding all the tiles together and adding a vertical scrollbar), accessible via single interface. Kinda like Planet++
This would allow a player to "micro" everything on a "macro" level. The sector would be able to build multiple buildings at once (or maybe one but couple of times faster), and have increased pop growth across the entire sector, but only would grow one pop at a time. The spaceport tab would have all the spaceports, not just one, for purpose of module upgrades, but it would have a singular queue out of which it would build.
This woudl eliminate need for sector AI, though one could still be provided via on/off interface plus some options.
This would allow a player to "micro" everything on a "macro" level. The sector would be able to build multiple buildings at once (or maybe one but couple of times faster), and have increased pop growth across the entire sector, but only would grow one pop at a time. The spaceport tab would have all the spaceports, not just one, for purpose of module upgrades, but it would have a singular queue out of which it would build.
This woudl eliminate need for sector AI, though one could still be provided via on/off interface plus some options.
The point is I don't want to micro all those tiles. Would a better interface help? Sure, but I still don't want to be microing every planet tile 100 years into the game.
You know what would actually help?
- Allowing us to build the max available rank of a building from the start.
- A small light/dot notification on a planet in the outliner if a pop is not "unemployed" but is only working a building-less tile bonus
I usually leave redevelopment on, mostly because the AI used to build tons of farms at the start of colonization and then replace them during later stages. But that seems to no longer be the case. In fact, as of right now, my experience has been that sectors vastly under-value farms at initial colonization and build them as some of the last stuff they construct. So it's probably not necessary to enable any more, even if you're giving over the planet day zero of its colonization.
Edit: What I don't get is why they don't just force the normal AI to respect tile yields. Yes it'd be less efficient than a player can be, but it'd still be WAY better than the current mess. The Sectors can at least build planets that work, even if they're not idealized all-one-type-no-spare-food planets. The AI can't even seem to manage that much for itself at the moment.
You mention food/planet twice in your post so I'll just say: You know food is empire-wide now right?
I'm aware, yes. Which is the only reason my planets didn't end up starving in my last game. There were several points during my second wave of colonization where I was down to -5 food because the sector controlling the planets flatly would not build farms and I had to add some to my main sector planets to make up for that. Infact, I lost my first Fanatical Purifiers run due to colonizing five new planets and having the AI completely fall apart and ending up stuck at -13 food with no easy way to fix it because the sector refused to develop more food to feed the people living there.
I experienced the opposite in 1.5.1. The AI simply refused to stop building farms, even on 2 mineral/1 food tiles while the empire was at 5000+400 food. =\
Weird. I definitely haven't seen it do that, though it does always end up heavily positive on food eventually. My experience has thus far been that sectors seem to build farms last, like it has some kind of aversion to stabilizing its food costs before it goes on to building other things. It's not really a noticeable problem past the really early game, since even just three planets with a few tiles dedicated to farms can reach enough positive that it doesn't matter. But it's gotten me into serious trouble a couple of times when I tried to colonize a large number of planets at once in the later part of the early game, before all my core worlds were completely built up with surplus food.
What sectors build is determined by the need of your empire, or at least that's what I saw. I was considering making a video showing what the sectors will build with given settings (balanced, etc...), so I set myself up with nothing but farms on the homeworld to maximize growth and let the sector get to work. Since I had a negative energy income, all the sector wanted to do was pump out power plant after power plant—on nineteen tiles it ended up with a planetary capital, a frontier hospital, three mineral buildings, and fourteen power buildings. Never mind that its tax rate was at zero, it wanted to help!
When I re-ran the test with a more balanced homeworld, I got the planetary capital, a frontier hospital, three farms, a mineral processing plant, a slave processing facility, a unity building, a lab, and ten power buildings. I was only getting +17 energy at that point, so maybe it still wanted to help boost that total... I ran that test after cheating to fully populate the world and giving the AI instant_build, so I'm not sure if that would have affected its final build or not.
When I re-ran the test with a more balanced homeworld, I got the planetary capital, a frontier hospital, three farms, a mineral processing plant, a slave processing facility, a unity building, a lab, and ten power buildings. I was only getting +17 energy at that point, so maybe it still wanted to help boost that total... I ran that test after cheating to fully populate the world and giving the AI instant_build, so I'm not sure if that would have affected its final build or not.
Based on my experience, I would have guessed that it would build things to try and bring its sector energy/minerals into the positive, then focus on filling out extra stuff that helps your empire. At least that's been pretty consistent with my sectors. They focus on building buildings which increase their energy and minerals first, along with starports, then food and orbitals later on. But I'm not sure how accurate that is, as it's purely anecdotal evidence from a couple of games I've played recently.
But I'd totally believe that they have some system set up to check what your total resource production is and try to stabilize it all. It just seems to put that at a lower priority in my experience than starting its own economy rolling, which also makes sense from a practical standpoint.
But I'd totally believe that they have some system set up to check what your total resource production is and try to stabilize it all. It just seems to put that at a lower priority in my experience than starting its own economy rolling, which also makes sense from a practical standpoint.
I am still fairly certain something like this could fix the issue in the buildings file (pay attention to the AI weight section near the end of each building):
This does not account for everything, as the decision is not entirely based on the weights in the building files alone, as can be seen with the factors in the defines file, but the ability to effect the issue here should be evident with the recent fix to Betharian and Xeno Pet tiles being fixed in these files by PDS themselves in 1.6
The amount of CPU strain this would cause is yet to be determined.
Code:
building_basic_mine = {
tags = { mining_network }
base_buildtime = @b0time
potential = {
planet = {
NOT = { is_planet_class = pc_habitat }
}
}
cost = {
minerals = @b0cost
}
produced_resources = {
minerals = @b0effect
}
planet_modifier = { #SM Additions
pop_happiness = -0.01 #Worst job ever
}
required_resources = {
energy = @b0upkeep
}
upgrades = {
building_mining_network_1
}
potential = {
custom_tooltip = {
text = "requires_shelter"
planet = {
has_building = "building_colony_shelter"
}
}
planet = {
NOT = { is_planet_class = pc_habitat }
}
}
ai_allow = {
NOR = {
has_resource = {
type = sr_betharian
amount > 0
}
has_resource = {
type = sr_alien_pets
amount > 0
}
}
}
ai_weight = { #<-------------------------------------------- THIS LOOKS IMPORTANT
factor = 1 #<------------- ALL TIER 1 AND 2 BUILDINGS ARE WEIGHTED AT 1 AND THUS ARE EASILY INTERCHANGED VIA REDEVELOPMENT THEORETICALLY
modifier = {
factor = 2 #<----------------------------------- THIS WAS NEVER HERE SHOULD PRIORITIZE BUILDING ON TILES WITH UNEMPLOYED POPS
has_grown_pop = yes
has_building = no
}
}
}
building_mining_network_1 = {
tags = { mining_network }
base_buildtime = @b1time
potential = {
planet = {
NOT = { is_planet_class = pc_habitat }
}
}
cost = {
minerals = @b1cost
}
produced_resources = {
minerals = @b1effect
}
planet_modifier = { #SM Additions
pop_happiness = -0.02 #Worst job ever
}
required_resources = {
energy = @b1upkeep
}
upgrades = {
building_mining_network_2
}
prerequisites = {
"tech_basic_industry"
}
allow = {
custom_tooltip = {
text = "requires_building_capital_1"
planet = {
OR = {
has_building = "building_capital_1"
has_building = "building_capital_2"
has_building = "building_capital_3"
}
}
}
}
ai_allow = {
NOR = {
has_resource = {
type = sr_betharian
amount > 0
}
has_resource = {
type = sr_alien_pets
amount > 0
}
}
}
ai_weight = { #<---------------------------------------------------- THIS LOOKS IMPORTANT
factor = 1 #<--------- CHANGE THIS TO 2 AND THEY SHOULD DEVELOP TIER 1 BUILDINGS LESS THEORETICALLY
modifier = { #<------------------- THIS MODIFIER SHOULD NOT BE NEEDED ON TIERS AFTER THIS ONE AS THIS IS THE FIRST TIER FOR COLONIES PAST SHIP SHELTER
factor = 7 #<------------------- THIS SHOULD MATCH OR EXCEED END LEVEL BUILDING TIER FACTOR
has_grown_pop = yes
has_building = no
}
}
}
building_mining_network_2 = {
tags = { mining_network }
base_buildtime = @b2time
is_listed = no
potential = {
planet = {
NOT = { is_planet_class = pc_habitat }
}
}
cost = {
minerals = @b2cost
}
allow = {
custom_tooltip = {
text = "requires_building_capital_1"
planet = {
OR = {
has_building = "building_capital_1"
has_building = "building_capital_2"
has_building = "building_capital_3"
}
}
}
}
produced_resources = {
minerals = @b2effect
}
planet_modifier = { #SM Additions
pop_happiness = -0.03 #Worst job ever
}
required_resources = {
energy = @b2upkeep
}
upgrades = {
building_mining_network_3
}
prerequisites = {
"tech_mining_network_2"
}
ai_weight = { #<------------------------------------ THIS WAS NEVER HERE AT ALL FOR THIS TIER
factor = 3 #<--- FACTOR SHOULD PROBABLY INCREASE WITH EACH TIER TO AVOID REDEVELOPMENT
}
}
building_mining_network_3 = {
tags = { mining_network }
base_buildtime = @b3time
is_listed = no
potential = {
planet = {
NOT = { is_planet_class = pc_habitat }
}
}
cost = {
minerals = @b3cost
}
allow = {
custom_tooltip = {
text = "requires_building_capital_2"
planet = {
OR = {
has_building = "building_capital_2"
has_building = "building_capital_3"
}
}
}
}
produced_resources = {
minerals = @b3effect
}
planet_modifier = { #SM Additions
pop_happiness = -0.04 #Worst job ever
}
required_resources = {
energy = @b3upkeep
}
upgrades = {
building_mining_network_4
}
prerequisites = {
"tech_mining_network_3"
}
ai_weight = { #<--------------- THIS WAS NEVER HERE AT ALL FOR THIS TIER
factor = 4 #<--- FACTOR SHOULD PROBABLY INCREASE WITH EACH TIER TO AVOID REDEVLOPMENT
}
}
building_mining_network_4 = {
tags = { mining_network }
base_buildtime = @b4time
is_listed = no
potential = {
planet = {
NOT = { is_planet_class = pc_habitat }
}
}
cost = {
minerals = @b4cost
}
allow = {
custom_tooltip = {
text = "requires_building_capital_2"
planet = {
OR = {
has_building = "building_capital_2"
has_building = "building_capital_3"
}
}
}
}
produced_resources = {
minerals = @b4effect
}
planet_modifier = { #SM Additions
pop_happiness = -0.05 #Worst job ever
}
required_resources = {
energy = @b4upkeep
}
upgrades = {
building_mining_network_5
}
prerequisites = {
"tech_mining_network_4"
}
ai_weight = { #<------------------ THIS WAS NEVER HERE AT ALL FOR THIS TIER
factor = 5 #<--------- FACTOR SHOULD PROBABLY INCREASE WITH EACH TIER TO AVOID REDEVLOPMENT
}
}
building_mining_network_5 = {
tags = { mining_network }
base_buildtime = @b5time
is_listed = no
cost = {
minerals = @b5cost
}
potential = {
planet = {
is_capital = yes
NOT = { is_planet_class = pc_habitat }
}
}
destroy_if = {
planet = { is_capital = no }
}
allow = {
custom_tooltip = {
text = "requires_building_capital_3"
planet = {
has_building = "building_capital_3"
}
}
}
produced_resources = {
minerals = @b5effect
}
planet_modifier = { #SM Additions
pop_happiness = -0.06 #Worst job ever
}
required_resources = {
energy = @b5upkeep
}
prerequisites = {
"tech_mining_network_4"
}
ai_weight = { #<-------------------------------------- THIS WAS NEVER HERE AT ALL FOR THIS TIER
factor = 6 #<------------------------------- FACTOR SHOULD PROBABLY INCREASE WITH EACH TIER TO AVOID REDEVLOPMENT
}
}This does not account for everything, as the decision is not entirely based on the weights in the building files alone, as can be seen with the factors in the defines file, but the ability to effect the issue here should be evident with the recent fix to Betharian and Xeno Pet tiles being fixed in these files by PDS themselves in 1.6
The amount of CPU strain this would cause is yet to be determined.
Last edited:
The added tile scope should not be needed in the buildings file as that is the default scope there.
This is what a lot of people aren't understanding. We're not asking for a perfect AI and in fact it might be because they tried to get too complicated that it went wrong. Right now, there are so many very basic ways, without getting too complicated, that the AI could improve on it's current performance. What it does right now is almost the worst things it could possibly do, in some cases it might even be better for it not to build at all than what it currently does, with all the resources it wastes. As you say, something as simple as just building the tile resource in the order of currently needed would be better than the current, and would be acceptable even if it isn't perfect.
We talking ideas or actual implementation in terms of what would be complicated? If the latter, a sounding board on my idea would be nice. I do not have time to test it right now, but it might work. If not, we could re-think the implementation and maybe work with AI_ALLOW = instead... or both?
From what I can see in the building files there is a decent chance for there to be a relatively easy solution to this. How much CPU strain it will add however is yet to be determined.
Thx for the video but 1.6 broke the whole game so it's not surprising to see the same result. Keep making such videos for future patches please, there are really interesting to see !
@Weedes; Your Code will probably help. There's a mod somewhere on Da Staem that implements some changes to that building file, and initial reports I've seen is that it does improve the AI's ability to construct the correct buildings. There's also the Shanty Town mod which helps by auto-placing 'shanty' versions of buildings which the AI will upgrade instead of redeveloping, usually.
The thing that may cause problems is that there's nothing in the buildings file which deals with priorities, indicating that some AI Logic is either in another file, or hard-coded into the Engine, probably the latter. That's something you might not be able to fix.
Now that I have some code to look at, the quick check to slightly increase weighting of an existing building should look like the following, replacing the has_building entry with the name of the building in question;
You could probably just implement this for the first tier, since you've also got a scaling weight for later Tiers. The risk might be that your scaling weight might prevent the AI from ever re-developing, even when it should, and when it wants to. But in my personal experience I don't think I've ever actually changed a building once it was there except for gradually replacing existing buildings with newly-unlocked Empire-Wide Modifier Buildings and The Power Hub.
Speaking of the Power-Hub, it should probably have a blanket increase for any tile with power, combined with a factor = 0 modifier if the planet already has one, which should be the following code;
There are almost assuredly ways that you could make the AI smarter just by modifying this file.
The thing that may cause problems is that there's nothing in the buildings file which deals with priorities, indicating that some AI Logic is either in another file, or hard-coded into the Engine, probably the latter. That's something you might not be able to fix.
Now that I have some code to look at, the quick check to slightly increase weighting of an existing building should look like the following, replacing the has_building entry with the name of the building in question;
Code:
modifier = {
factor = 1.1
has_building = building_mining_network_1
}You could probably just implement this for the first tier, since you've also got a scaling weight for later Tiers. The risk might be that your scaling weight might prevent the AI from ever re-developing, even when it should, and when it wants to. But in my personal experience I don't think I've ever actually changed a building once it was there except for gradually replacing existing buildings with newly-unlocked Empire-Wide Modifier Buildings and The Power Hub.
Speaking of the Power-Hub, it should probably have a blanket increase for any tile with power, combined with a factor = 0 modifier if the planet already has one, which should be the following code;
Code:
modifier = { factor = 2 has_resource = { type = energy amount > 0 } }
modifier = {
factor = 0
OR = {
planet = { has_building = "power_hub_1" }
planet = { has_building = "power_hub_2" }
}
}There are almost assuredly ways that you could make the AI smarter just by modifying this file.