Stellaris Dev Diary #385 - AI Benchmarks

[rubert]

Played that before, it's not wishable. You just end up abusing Ai behaviors. It'll teach how to play genocidals by dodging fleet and aiming for the worlds well, but it's kind of a chore (especially with the lags back then). Starnet Ai really illustrates that.

We need smart ai, not 10+ year0 crisis.

I am not arguing against smarter and better ai. My original comment was to the comment where you mentioned GA ai overflowing with resources. If the AI has resources the it should use them because it is a smart thing to do. If that breaks other parts of the game, like Fallen Empires getting killed too early then they should be buffed.

If it makes game too difficult or the player feels like they have to use annoying tricks they can always reduce the difficulty.

Even in non-Admiral difficulties I saw the biggest AIs to snack on Fallen Empires, sometimes well before mid game year. Power level of normal empires (both human and AI) have increased over the years due the DLC and Fallen empires really need updates, or very least simple buffs to number ships they receive based on the game difficulty setting.

 

[banachraum]

Please just get rid of this cheating difficulty settings. There have been mods showing you can let the Ai play optimal or near optimal. Please try to archieve this and than maybe nerv it and/or use Ai personas if there have to be difficulty settings. The game is meaningless if the AI is no challenge. I agree with the other posters that AI should be really top priority instead of multiplayer. Also please take balance more serious, its horrible that there are still so many bad/useless civics etc... at least try to bring them all to a similar level. I know its not possible, but I would really like if instead of adding system upon system, making the game even more complex and therefor unbalanced,bugged and the AI not working, you would take maybe a year to fix and balance all systems already present. The game is not in a state for even more content,systems.

 

[Trigaea]

Can't you gather the data directly from the game app? Wouldn't it provide a much better sample than asking people on the forum?

Having different types of data sources is good.

You don’t need it for a question that has a simple and clear answer like “What’s your favorite ice cream flavor?” or “What ethos does the player base favor?”. However, it’s good for questions might have complicated or unexpected answers or where a seemingly simple question needs a more thorough answer.

In this case, it’s likely they collect these data points, but the aggregate data of the entire playerbase might not he helpful. Even a subset, such as data from players with >100 hours might not be helpful. Based on the phrasing of the question, it sounds like they’re partially interested in what “feels” like good benchmarks (or the vibe depending on if you were born before or after 2000).

But also, they could just be asking for engagement purposes + to anecdotally confirm the data they already have.

 

[IMyrmidon]

Having different types of data sources is good.

You don’t need it for a question that has a simple and clear answer like “What’s your favorite ice cream flavor?” or “What ethos does the player base favor?”. However, it’s good for questions might have complicated or unexpected answers or where a seemingly simple question needs a more thorough answer.

Now I'm curious about the ethos thing. For me, it's definitely xenophile or gestalt consciousness.

 

[Calvax]

Now I'm curious about the ethos thing. For me, it's definitely xenophile or gestalt consciousness.

Egalitarian and/or xenophile are probably my most played.

I like the power fantasy of creating an empire that makes life better for people.

 

[Neal?]

I wonder if it would be possible to have a scaling reduction to research based on the difference in research income. So it's not that the techs cost MORE for specializing, but your specialization costs LESS. Now that research is split up again, and there are more society weapons than ever. I think it would be really neat to lean into science specializations. So maybe something like a 1% tech cost reduction for every 15% higher your research income is over othe research incomes with a cap of like 25%

 

[grommile]

You know, a person doesn't have to be a developer to have every right to disdain both the Paradox testers and their management.

This isn't nuclear power or aviation. QA do not have veto power over releases, so don't blame QA for management's decisions.

 

[relytor]

TLDR; specializing planets is more important in this update than ever before, and the AI is completely failing at that.

First, I want to say that I feel this is the biggest step forward for the game in general than anything else that has come out since 2.2. I commend you on that! There are several mod ideas I've had that just didn't work that well with the old system that I can now see making work how I want. While it isn't polished yet, this is the first update in two years I've actually not got bored with after a few days, and gone back to EU4 or CK3.

I'd also like to say I am not a "map painter". I usually go to war for a purpose; I'm lacking a resource I can take from my neighbor, I need to stop them from getting too strong, etc. I very rarely go to war just to take more land. If I am in a "map painter" mood, that's when I will play a genocidal. But I will then also usually play on a medium map, whereas normally I much prefer the largest map.

Now for the AI; I think a big problem is that the AI is trying to build toward targets over time, instead of trying to optimize their current situation. In this update, I find myself deciding what to build based on how many Civilians I have. If I need a resource, but all my planets already have open jobs and no Civilians, I'll retool what I currently have: replacing Mining districts to Farming, replacing a Factory with a Foundry, or changing the zoning from Archives to Administrative. Only if it is just a short-term need will I use the job sliders to just prioritize different jobs. If I have free Civilians, then I will build something new. But, I keep planets specialized, and will wait for them to migrate to the specialized planet I need them on, or move them directly if it is urgent and my empire/ethics allows it. With the new update, specializing is especially important with the District Specializations and all the new buildings, as it requires three buildings to optimize any specialisation: the bonus production building, the job efficiency building, and the job upkeep reduction building. Then, I will usually build at least one of each of the rural districts on each planet so I can get access to those building slots. In the early game, I'll usually just fill those rural building slots with Resource Silos until I get better use for them.

When the AI does do some specialization, it also seems to have a difficult time deciding what specialization a planet should have. You'll see them use a planet with bonus food production as their mainly Foundry planet, and have another planet with a lot of farming district slots but no bonus as their big food producer. Which I find interesting, because the game is usually pretty good at deciding Planetary Designations that are appropriate much better than the AI does at actually deciding what to build on a planet.

The old "meta" of taking as many planets as possible, even low habitability ones, because every planet meant more overall pop growth seems to have been "solved" in this update. If you don't have the population to build up a new world, it sits at low population forever with the new "low population" growth modifier. I personally think that is a huge win for the game, where prior you had to make a big land grab at the very beginning or you couldn't keep up. One of my first games in this update, I felt lucky because I had 5 high-habitability planets near my start, and I colonized all of them very early. And they all sat with very low populations forever, acting more as a drain on my homeworlds resources than contributing to my empire. I was Egalitarian, so I didn't have forced pop movement, so just had to wait for natural migration and growth. This is how I believe colonization should feel! It should be an expensive, slow start that takes a long time to have returns! Once I learned to start with just the two Guaranteed habitable planets and allow them to develop until they can contribute to further colonization does settling new worlds now make sense. Basically, my new rule tends to be not to settle a planet unless I several hundred population ready to migrate there.

As an exclusively Single Player user, I don't have expectations for reaching certain thresholds at different times in the game, as it depends on my tech and situation. The one thing I find unique about Stellaris over all the other Pdx GSGs is that it isn't the asymmetric start of the others. In EU4, I will choose which nation I play based on what kind of game I'm in the mood for. I want something laid back and just to concentrate on my economy, I'll choose a nation that I know the neighbors will allow me to do that. In Stellaris, you can go in expecting this game to be a certain way, but your first contact can completely change that. I want a low-key, tall, economy game, but I don't know if my first contact is going to be a Devouring Swarm or a peaceful Megacorp. As such, Stellaris requires a much more dynamic playstyle. But, within that dynamism, there still has to be certain rules you have to follow. If I want to remain peaceful, I have to keep my fleet power at least equivalent to all my neighbors. If I want to be aggressive, I want to keep my fleet power superior to my neighbors. Sometimes that requires making more soldier jobs and building Anchorage starbases, sometimes that's not required. Basically, my goal thresholds are dynamic based on my neighbors. If I'm Superior to my neighbors in tech, but falling behind in fleet or economy, I might retool a bit to tone down research in favor of alloys, for instance. But, I'm always reacting to the game world around me, not just building toward the same goals every game.

But even so, my early game goals are always the same; to get my specialised planets setup: a Research, a Foundry, a Factory, a Food, a Trade/Energy. Usually you get enough space minerals, that I don't concentrate on a mining planet until mid-game. I also tend to leave my Capital as a dynamic jack-of-all-trades that usually doesn't get specialized until mid-game, when it usually become either my Unity or Trade center. Until then, it is usually my Capital that gets retooled as needed, fulfilling whatever shortcomings my current specialized planets are having until they get fully realized.

Also early game, of course, I concentrate on establishing my borders. Send out my military ships to find the nearest choke points, and see if any empire has spawned near me. Then I can prioritize taking those chokepoints and establishing a strong early-game boundary. Unless I'm playing aggressive, which usually means genocidal, in which case I am specifically looking for that first weak neighbor to take out as soon as possible, as I'd rather spend the resources taking their "developed" homeworld than starting my own from scratch at this early point.

Currently, I'm just seeing the AI do none of that. The state of the AI determines the difficulty I play on. 3.12, the last version I played more than a few hours, I found that the AI had gotten quite a bit better than previously, and usually found Commodore (Mid-game) as a good difficulty. This update I've been playing with Grand Admiral (Mid-game), and still usually far outpacing the AI. The biggest change you could make for them is to prioritize specialization, and then from there make tweaks. But I much prefer an AI that plays well over one that just gets a bunch of bonuses.

 

[Trigaea]

Now I'm curious about the ethos thing. For me, it's definitely xenophile or gestalt consciousness.

Think the last time Eladrin mentioned it, he said xenophile was the most played.

Maybe he’ll see this message and grace us with some more factoids.

 

[glee8e]

I think your first priority is to fix planetary automation. Before 4.0 it was working correctly most of the time, so AI tend to build planets the way I did. Add some slight buffs and it can be up to a challenge. Even if it doesnt get any, it can still stand on its own feet. Now the planetary automation seems to just not work at all. There are always huge unemployment on every AI planet. Even when I tried it myself on my own planets, it never builds anything except the first 10 years the planet is colonized, or when I'm in some resource shortage (in that event it builds so much that it makes another resource in deficit by draining too much pop).

AI benchmarking or anything doesn't matter here. It's just a basic requirement for AI to work at all.

 

[SarahGabriella]

When i play a peaceful empire, i strive to built or join a federation, built alliances for defense purposes, so i have fleet build up as a secondary objective. The primary objective is to just tend to my people. I expect a peaceful AI to be similar to that.

 

[TeHb Z Shadow]

“please don’t collapse in an economic death spiral” - in 4.0.X, I still often see AI building a colony ship, but then refusing to actually use it.

Like I met The Chosen, depopulated a few of their planets including the homeword (but not the other 2 planets in the homeworld system), and they would never recolonize their homeworld, despite having a colony ship in the system for decades.

Later I wiped them out, and to this day, nobody wants to colonize the now empty cluster, but I have seen science ships scanning them, so they know how to use a wormhole. I wonder why AI does not see a system with 3 gaia worlds as something worth colonizing.

AI will never settle further than 2 jumps from their border. Who controlled another side of wormhole, AI or you?

> I’m interested to hear what you all strive for.
> How much research and alloy production do you try to have 10 years, 30 years, 100 years, and when the end-game crisis comes calling? (Include your preferred difficulty settings and galaxy sizes as well if you could, as well as if you change any other important settings like tech costs.)

Can't you gather the data directly from the game app? Wouldn't it provide a much better sample than asking people on the forum?
A genuine question, no pressure.

Okey, lets imagine they decided to check my last 10 games using data:
Devs:
-Ok, lets see. Player has 1 fully upgraded merc enclave fleet, he is on same fleet power as some other AI. Wow, we made a perfect balance!

Me:
-Oh god, AI is so stupid, i see no sense to go any military power, AI will never beat my single Enclave fleet at this stage/my federation members. So i will just invest everything in economy.

Data says it's all fine, feedback says AI should be better.

 

[SirBlackAxe]

Here are some datapoints from my current WIlderness (note: bioships, Natural Neural Network, Elevational Desires) game on the new 4.0 economy. Normally I play on Grand Admiral, midgame scaling, medium galaxy, 2275 midgame / 2350 endgame / 2400 victory, although this game I was experimenting with using Difficulty Adjusted AI modifiers on unscaling Commodore. I'm not sure how useful the atypical wilderness economy is for benchmarking, though.

• 15 years: 3 colonies, 151 empire size
  • food = 237 +80
  • alloys = 227 +12 (monthly trade +7)
  • unity = 1.2k +66 (1 tree completed, 1/5th 2nd)
  • research = +171
  • fleet = 21/23
• 29 years: 4 colonies, 192 empire size
  • food = 6.6k +94
  • alloys = 69 +10 (including monthly market purchase - currently building two moltgut specializations on my first alloy world though so this is about to spike)
  • unity = 5.5k +191 (2 tradition tree complete)
  • research = +263
  • fleet = 37/60
• 84 years: 7 colonies, 314 empire size
  • food = 83k +1081
  • alloys = 57k +467 (this feels high)
  • unity = 43k +2.8k (6 tradition trees complete, all 7 colonies ascended to tier 4. Note that I got the Propagandasphere curator relic) (this feels very high)
  • research = +4.6k (this feels very high)
  • fleet = 293/672 (Note I have the food/alloys to build a much larger fleet, but I'm already maxing power projection and all AIs except one inferior are pathetic so I don't need to)

Benchmarks aside, I think AI should aim to have a minimum set of city specializations between its starting worlds. Between its capital and 2 guaranteed habitables (or other good habitability colonies) it should have one each of a Heavy Industry, Research Enclave, and Administrative Hub / Spiritual Enclave before it starts building other or repeated specializations. Use buildings and the market for consumer goods until you've got a 4th planet or Colonial Centralization. Or maybe unaggressive personalities with ethics/civics that want to focus on unity/research should build a Mixed Industry instead of Heavy Industry? Which should go on the homeworld (swapping one of the starting specializations out so it's got a 3/4 focus) and which of the other two goes on the larger colony should be depend on ethics/civics/government, e.g. Militarists and empires with military civics/personalities/governments should swap their homeworld Archives for Heavy Industry, while Materialists / researcher civics should swap their homeworld Mixed Industry for a Research Enclave.

 

[Foxosaur]

Speaking of AI, I'm enjoying seeing the toast notification of when empires are embracing new or fanatic ethics. Whether the behaviour is new I'm itself, I'm not sure.

 

[Koopatin]

i just try out a bit. and ignore min maxing. and because the poster above had such a nice list i will use the same.

grand admiral - scaling - to mid game (2325) - 800 systems

• 153 years: 44 colonies, 27,6k Pops, 1360 empire size
  • food = +79
  • energy = -81 (fleets at war - no dyson sphere yet)
  • minerals = +1,8k (no megastructure yet)
  • alloys = +910
  • consumer goods = -145
  • trade = +1,3 k
  • unity = +2,3k
  • research = +6k
  • fleet = 1176/1080

and this is my strongest enemy - devouring swarm AI. was stronger before our war but did split his fleets a bit much.

• 153 years: 47 colonies, 46,7k Pops, 805 empire size
  • food = +429
  • energy = 1,4k (fleets at war - no dyson sphere yet)
  • minerals = +2,1k (no megastructure yet)
  • alloys = +1,4k
  • consumer goods = 148
  • trade = +442
  • unity = +856
  • research = +2k
  • fleet = 1144/1953
• reminder above AI numbers are with boni from difficulty - after tag switch they lose that only after month tick

Most of this Swarms Planets do have between 0 and 2 buildings. Some have specialisations without additional buildings. some have not even that.

 

[jedavis]

The AI has low pops to economy ratios(based on GC numbers with the bonuses removed). Meanwhile, players can reach 5-8x pops to economy ratio

This sounds like an interesting metric but I'm not sure exactly what you mean. Are you talking about like, taking economy score and dividing it by number of pops? I found that my economy score has been grossly inflated by the huge production of strategic resources from metallurgists with the new refining system, without being representative of my "real" economy (having huge surpluses of motes that I can hardly even sell for much doesn't help). So I might be a little cautious about economy score - but maybe that isn't even what you're using?

 

[JagoTheFanaticalPurifier]

Please fix opinion modifier for xenophobes not calculating. Xenophobes are too cordial!

 

[tarkain]

As I mentioned last week, with multiplayer stability largely handled, AI is one of our next focuses. Today I want to talk about AI benchmarks, and have a discussion with you about how we should measure “success”.

Forgive me if I'm wrong, but I don't know if you're talking about stability regarding crashes or other issues. However, if you include desynchronization problems, forgive me, but it's absolutely not fixed (unless you need to restart with a new game, in which case I apologize). Just last night, during a multiplayer session with a friend, we experienced a ton of desynchronization issues, including "resources" desync alerts that I thought had been corrected.


Again, I repeat: I'm speaking out of exasperation because Stellaris is a wonderful game, just like your other games. I only want to play Stellaris or CK3, but my teammate and I quickly lose patience due to desync and crash issues. However, if creating a new game would be enough to resolve these problems, then please forgive me for this message

 

[Rasputin92xl]

Hi everyone!

The 4.0.13 update released today with the following changes:

Spoiler: 4.0.13 Release Notes

Stellaris 4.0.13 Patch​

Improvements​

• Behemoth Fury is now available to Wilderness Empires.
• Improved tooltips for the following civics:
  • Functional Architecture/Constructobot
  • Environmentalist
  • Astro-mining Drones
  • Maintenance Protocols
  • Ascensionists
  • Augmentation Bazaars
  • Brand Loyalty
  • Death Cults
  • Dimensional Worship

Balance​

• Mutagenic Habitability now counts all planet types as ideal for upgrading Gaiaseeders
• Dramatically increased the draw chance for the Mineral Purification, Global Energy Management and Food Processing technologies
• Rebalanced the Pleasure Seeker civics to transform Civilians into Hedonists
• Logistic Drones are now Complex Drones not Menial Drones

Bugfix​

• Fixed invaded pre-ftls not becoming biotrophies
• People once more die when they are put in the Lathe
• Bio-Swarmer missiles can now be used by all biological ships with medium weapon slots (including defensive platforms)
• Pops that are being pampered will now be forcibly switched to the correct living standard
• Replacing a district specialisation no longer destroys CyberCreed buildings that should be kept
• Corrected a tooltip bug where a planet would display itself as a possible migration target.
• Fixed capitalisation for resources in trade policies
• Updated assorted modifiers that still referred to Clerks
• A Trade deficit now causes Job Efficiency and Empire Size issues
• Fixed the tooltip for the Polymelic trait
• Armies now protect 200 pops from raiding, not 2
• Blocked the Federation Code technology for some empires, for example homicidals. To draw the tech, the empire is also required to be in contact with someone they can form a federation with.
• Blocked the Development focus task Form a Federation for some empires, for example homicidals
• Added swaps for some empires, for example homicidals, for the Development focus rewards Federation Code, Xeno Diplomacy, and Xeno Relations
• Updated the Colony view tab mentioned in the hint of the focus task Enact a Planetary Decision to say Management
• CyberCreed pops with Ritualistic Implants can now colonise planets
• Fixed Recycled and Luxurious traits not applying to Roboticists
• Catalytic Processing Civic now lists correct information regarding job swap
• Cost for repairing orbital rings when you use bioships is now correctly calculated
• Gale Speed trait gained from Defeat no longer causes errors
• Fixed scope for LeaderShipSurvivalReason
• Fixed scope bug for ruler in leader_election_weight
• Fixed Worker Coop gaining Elite strata jobs in too many places and tidied up the civic tooltip
• Updated tooltip for Warrior Culture civic
• Added a pre-list colon to the Feudal Society civic's tooltip
• The everychanging stone can no longer cause artisans to have negative mineral upkeep
• Gave the Neural Chorus advanced authority the pop growth speed modifier that had accidentally been assigned to Memory Aggregator
• The Planetary Supercomputer no longer has an empire cap of 1
• The Research Institute/Planetary Supercomputer no longer give scientist capacity
• Added dashes to Traits tooltips and list items
• Fixed trigger logic for criminal syndicates and federations
• Fixed Offspring Bioships not being visible in game
• Fixed Offspring Bioships not being labelled as non-offspring ships in the ship designer
• The Machine Uprising will no longer spawn 100 machine pops for every 1 missing housing. However the pop-rework seems to have handled 6 million machine pops okay.
• Stopped removing occupation armies for bombarded and invaded planets on savegame load
• Repairing ruined buildings in zones is now always possible.

Performance​

• Flattened pop job modifier node into planet one
• Made clearing modifiers a fire and forget job

Stability​

• Fixed a possible OOS when a player leaves the game.
• Fix CTD when generating a Cosmic Storm mesh.
• Fixed a random freeze when loading save with stations containing multiple defence platforms.

We expect the 4.0.14 release will be next week (probably on Tuesday), and is expected to include some fixes to a few infinite loops and some select balance changes (like splitting up Enforcers and Telepaths again). It will be a short work week here in Sweden, so it’s likely to be the only update of the week.

As I mentioned last week, with multiplayer stability largely handled, AI is one of our next focuses. Today I want to talk about AI benchmarks, and have a discussion with you about how we should measure “success”.

What Makes a Good AI?​

The AI in Stellaris has always been designed as very reactive, and AI personality has a massive impact on their behavior. Our goal is for our AI empires to feel like actors in the galactic play - acting in a manner consistent with their Origins, Authorities, Civics, and Ethics rather than always picking the “meta” play.

They do still need to put up a bit of a challenge though, especially at higher difficulties.

The first economic goal we make for our AI is “please don’t collapse in an economic death spiral”, and it’s actually far better at that in 4.0 than it was in 3.x. The current AI does NOT meet the second “provide an adequate challenge” goal though.

One of the fundamental tools we have for our AIs are resource targets in their economic plans. They’ll strive to reach those targets, and many of these are set as “scaling” - if they meet the target, they’ll raise the target the next month. This attempts to ensure that they’ll keep thirsting for ever larger research and alloy numbers (or food if they use bioships!) as is appropriate. This is one of the tools we also use to make them exhibit their ethics - Materialists scale their Research targets faster than other empires, so they’ll inherently be more likely to build more Research specializations, while Spiritualists are more likely to have a lot of Unity specializations.

Ironically, improving AI tends to consume any benefits we carved out through performance improvements. The stronger the AI, the more stuff they have - fleets, colonies, and so on.

Benchmarking​

One way to decide whether or not the AI is performing up to expectations is through benchmarking - what kind of fleet power, alloy generation, and research generation should they have by 2230, 2250, 2300, and so on? Around what year should they hit 10k fleet power?

Then there come questions around whether the benchmarks should differ based on personality type. Should it be different if they’re Democratic Crusaders vs. Peaceful Traders? Or does differentiating them there make the friendlier empires too weak?

I’ve got my own set of benchmarks that come from running 3.14 and from the multiplayer community, and in general, I’m okay with Grand Admiral being significantly harder than it was in 3.14. but I’m interested to hear what you all strive for.

How much research and alloy production do you try to have 10 years, 30 years, 100 years, and when the end-game crisis comes calling? (Include your preferred difficulty settings and galaxy sizes as well if you could, as well as if you change any other important settings like tech costs.)

What’s Next?​

We’re going to continue with 4.0 post release support.

Since the next two weeks are both short weeks in Sweden, our next Stellaris Dev Diary will be June 12th. (You’ll be hearing from me in patch notes in the meantime though.)

Hello I love this game.. But I can not say same about this update 4.0 ... I do not know what your testers was testing... I played 3 games in 4.0 together 170+ hours... In all games AI in later mid game/late game... start falling apart... when they get more advanced technologies... I was playing on Grand Admiral... and in late game AI was having milions planets... most of them mepty... 1-2 buildings... Starbases and Planets completly wrong building placement.... And more complicated race was... and more research and options they have... they get worst and worst... I was not see in 3 games one strong plant ship Country... AI have problem to handle thinks like Ecumenopolis and other advance planets... AI killing their income and possible progress be building stupid buildings... even Country with big territory and a lot of planets was in late game on Grand Admiral not doing much... AI in this 4.0 update is biggest problem... I hope you guys solve this soon... thanks

 

[Alpha_Centauri]

Hi everyone!

The 4.0.13 update released today with the following changes:

Spoiler: 4.0.13 Release Notes

Stellaris 4.0.13 Patch​

Improvements​

• Behemoth Fury is now available to Wilderness Empires.
• Improved tooltips for the following civics:
  • Functional Architecture/Constructobot
  • Environmentalist
  • Astro-mining Drones
  • Maintenance Protocols
  • Ascensionists
  • Augmentation Bazaars
  • Brand Loyalty
  • Death Cults
  • Dimensional Worship

Balance​

• Mutagenic Habitability now counts all planet types as ideal for upgrading Gaiaseeders
• Dramatically increased the draw chance for the Mineral Purification, Global Energy Management and Food Processing technologies
• Rebalanced the Pleasure Seeker civics to transform Civilians into Hedonists
• Logistic Drones are now Complex Drones not Menial Drones

Bugfix​

• Fixed invaded pre-ftls not becoming biotrophies
• People once more die when they are put in the Lathe
• Bio-Swarmer missiles can now be used by all biological ships with medium weapon slots (including defensive platforms)
• Pops that are being pampered will now be forcibly switched to the correct living standard
• Replacing a district specialisation no longer destroys CyberCreed buildings that should be kept
• Corrected a tooltip bug where a planet would display itself as a possible migration target.
• Fixed capitalisation for resources in trade policies
• Updated assorted modifiers that still referred to Clerks
• A Trade deficit now causes Job Efficiency and Empire Size issues
• Fixed the tooltip for the Polymelic trait
• Armies now protect 200 pops from raiding, not 2
• Blocked the Federation Code technology for some empires, for example homicidals. To draw the tech, the empire is also required to be in contact with someone they can form a federation with.
• Blocked the Development focus task Form a Federation for some empires, for example homicidals
• Added swaps for some empires, for example homicidals, for the Development focus rewards Federation Code, Xeno Diplomacy, and Xeno Relations
• Updated the Colony view tab mentioned in the hint of the focus task Enact a Planetary Decision to say Management
• CyberCreed pops with Ritualistic Implants can now colonise planets
• Fixed Recycled and Luxurious traits not applying to Roboticists
• Catalytic Processing Civic now lists correct information regarding job swap
• Cost for repairing orbital rings when you use bioships is now correctly calculated
• Gale Speed trait gained from Defeat no longer causes errors
• Fixed scope for LeaderShipSurvivalReason
• Fixed scope bug for ruler in leader_election_weight
• Fixed Worker Coop gaining Elite strata jobs in too many places and tidied up the civic tooltip
• Updated tooltip for Warrior Culture civic
• Added a pre-list colon to the Feudal Society civic's tooltip
• The everychanging stone can no longer cause artisans to have negative mineral upkeep
• Gave the Neural Chorus advanced authority the pop growth speed modifier that had accidentally been assigned to Memory Aggregator
• The Planetary Supercomputer no longer has an empire cap of 1
• The Research Institute/Planetary Supercomputer no longer give scientist capacity
• Added dashes to Traits tooltips and list items
• Fixed trigger logic for criminal syndicates and federations
• Fixed Offspring Bioships not being visible in game
• Fixed Offspring Bioships not being labelled as non-offspring ships in the ship designer
• The Machine Uprising will no longer spawn 100 machine pops for every 1 missing housing. However the pop-rework seems to have handled 6 million machine pops okay.
• Stopped removing occupation armies for bombarded and invaded planets on savegame load
• Repairing ruined buildings in zones is now always possible.

Performance​

• Flattened pop job modifier node into planet one
• Made clearing modifiers a fire and forget job

Stability​

• Fixed a possible OOS when a player leaves the game.
• Fix CTD when generating a Cosmic Storm mesh.
• Fixed a random freeze when loading save with stations containing multiple defence platforms.

We expect the 4.0.14 release will be next week (probably on Tuesday), and is expected to include some fixes to a few infinite loops and some select balance changes (like splitting up Enforcers and Telepaths again). It will be a short work week here in Sweden, so it’s likely to be the only update of the week.

As I mentioned last week, with multiplayer stability largely handled, AI is one of our next focuses. Today I want to talk about AI benchmarks, and have a discussion with you about how we should measure “success”.

What Makes a Good AI?​

The AI in Stellaris has always been designed as very reactive, and AI personality has a massive impact on their behavior. Our goal is for our AI empires to feel like actors in the galactic play - acting in a manner consistent with their Origins, Authorities, Civics, and Ethics rather than always picking the “meta” play.

They do still need to put up a bit of a challenge though, especially at higher difficulties.

The first economic goal we make for our AI is “please don’t collapse in an economic death spiral”, and it’s actually far better at that in 4.0 than it was in 3.x. The current AI does NOT meet the second “provide an adequate challenge” goal though.

One of the fundamental tools we have for our AIs are resource targets in their economic plans. They’ll strive to reach those targets, and many of these are set as “scaling” - if they meet the target, they’ll raise the target the next month. This attempts to ensure that they’ll keep thirsting for ever larger research and alloy numbers (or food if they use bioships!) as is appropriate. This is one of the tools we also use to make them exhibit their ethics - Materialists scale their Research targets faster than other empires, so they’ll inherently be more likely to build more Research specializations, while Spiritualists are more likely to have a lot of Unity specializations.

Ironically, improving AI tends to consume any benefits we carved out through performance improvements. The stronger the AI, the more stuff they have - fleets, colonies, and so on.

Benchmarking​

One way to decide whether or not the AI is performing up to expectations is through benchmarking - what kind of fleet power, alloy generation, and research generation should they have by 2230, 2250, 2300, and so on? Around what year should they hit 10k fleet power?

Then there come questions around whether the benchmarks should differ based on personality type. Should it be different if they’re Democratic Crusaders vs. Peaceful Traders? Or does differentiating them there make the friendlier empires too weak?

I’ve got my own set of benchmarks that come from running 3.14 and from the multiplayer community, and in general, I’m okay with Grand Admiral being significantly harder than it was in 3.14. but I’m interested to hear what you all strive for.

How much research and alloy production do you try to have 10 years, 30 years, 100 years, and when the end-game crisis comes calling? (Include your preferred difficulty settings and galaxy sizes as well if you could, as well as if you change any other important settings like tech costs.)

What’s Next?​

We’re going to continue with 4.0 post release support.

Since the next two weeks are both short weeks in Sweden, our next Stellaris Dev Diary will be June 12th. (You’ll be hearing from me in patch notes in the meantime though.)

I generally don't set any goals for myself. I just grow my economy in bits and pieces organically as I colonize and expand.