I've seen a lot about how the A.I. is terrible, but I haven't seen a lot of ideas on how Paradox might address how the A.I. works. So lets take a look at how the A.I. functions. Additional observations and ideas are welcome.
- Planetary specialization
- The A.I. is weighted to chose a planets specialization based on what modifiers the planet has, what max districts and current districts, as well as buildings. Notably research worlds are unaffected by the presence of research labs, and will instead be triggered only by planet modifiers.
- The important thing to note is that specialization does not appear to affect what the A.I. builds. While sector automation is supposed to use the colony automation files to determine what to build, the A.I. does not appear to use the same build order.
- Planetary Building
- The A.I. only has two criteria that determine what it will build; can it afford it, and it’s assigned weight.
- Weight varies from building to building, but it appears that the majority of weights are based on flat income values, or already existing iteration of the same building. For instance, the A.I. will only build 3 research labs on a planet normally. They may build 5, if they are behind on tech. They will also stop building more if they upgrade an existing lab. The A.I. will also build at least one factory per world, but may build more if their income reduces.
- An oddity with this system is that it doesn’t matter what the weight of building something is, if that item is cheaper than all other options. That’s why the first building the A.I. produces is the autochthon monument.
- Job oddities
- While the A.I. and the player use the same system to determine how jobs should be filled, the A.I. has a couple limiters in place that mess with their ability to fill jobs correctly. For instance, the A.I. will not fill researcher jobs if their stockpile of consumer goods is less than 500.
- What does all this mean?
- It means that the A.I. build unnecessary buildings early on, while at the same time being unable to leverage the specialization gameplay mechanic correctly. So the A.I.’s economy is rather slow to grow in comparison to a player. Though I will say that this approach does mean that an A.I.’s economy will tend to be more stable as a result.
- How to fix this? – Leverage planetary specialization
- Planetary specialization should not be chosen by the A.I. because of what a planet already has, but rather by what the A.I. wants the planet to become. Each world gets an initial designation of urban or rural. Then, based on economic needs, those worlds can later be further specialized. If the A.I. needs more alloys, it will change an urban worlds specialization to forge to will that need.Planetary build order can then either be determined by using the existing sector automation system, or tying the weights of buildings to planetary specialization.
- Additionally, the A.I. should be able to replace existing buildings and districts if the planetary specialization no longer matches them. So a forge world will get rid of competing specialist buildings, while an urban world would remove excess forges in favor of what the economy needs currently.
Fleet A.I. –
- Station upgrades
- How the A.I. determines where to upgrade stations is unknown. From early game observation, it seems the A.I. favors creating starbases over colonized worlds, rather than building initial defensive positions
- What we do know is how the A.I. specs their starbases. Combat modules and buildings can only be built at chokepoints, Non-Combat are built at non-chokepoint starbases. Exceptions that can be built anywhere are the hanger, crew quarters, nebula refinery, art college, and think tank.
- As to weights; Buildings with a unique requirement have a weight of 150, buildings that require modules have 200 as long as at least3 of those modules are present. Other non-combat buildings have a weight of 50, and combat buildings are 10. Modules are less uniform; shipyards are 50. Anchorages are 40 for the first 2, then 20. Trade or solar are 15. Weapons are 10, or 20 if it’s the civilization’s preferred weapon.
- Fleet movement
- When at peace; The A.I. appears to station its fleets primarily around stations. If the fleets are strong enough, and perhaps some threshold is fulfilled, the A.I. will move to attack hostile fauna.
- When war is declared, the A.I, will move their fleets to attack the nearest known threat. So the A.I. is not capable of taking a defensive position. The A.I. does appear to not attack enemies that pose to large of a threat. (In a recent game, I’ve held a fortified system for years without seeing the enemy)
- This behavior does currently cause a bug; if the A.I. does not see the enemies fleet, they may instead move to attack local fauna when war is declared. This may be because the A.I. is coded to attack known fleets first, the largest threat first, nearest threat first, or if the A.I. simply doesn’t have vision on the enemy stations causes a glitch.
- Add the “station specialization” concept to the A.I., allowing them to plan how they want to equip a new starbase.
- I think the A.I. placement of Stations could also be improved, and perhaps personalized. Currently the A.I.’s placement is sub-optimal. The only reason to be building starbases over planets is for trade collection, or blacksites. But early on the A.I. hasn’t expanded out far enough to need those. Early on, creating border forts at key chokepoints is much more important than collecting an extra 10 trade. Starbases are a limited resource, and the A.I. needs to be able to leverage them properly.
Source: Original link
© Post "Let’s talk about the A.I. – Brainstorming ideas." for game Stellaris.
Top 10 Most Anticipated Video Games of 2020
2020 will have something to satisfy classic and modern gamers alike. To be eligible for the list, the game must be confirmed for 2020, or there should be good reason to expect its release in that year. Therefore, upcoming games with a mere announcement and no discernible release date will not be included.
Top 15 NEW Games of 2020 [FIRST HALF]
2020 has a ton to look forward to...in the video gaming world. Here are fifteen games we're looking forward to in the first half of 2020.