Hey all,
New version inbound. I am exhausted so I'm just gonna drop the changelog and go. Many thanks to @Enginseer, @axatin, @seroperson, @Putmalk, @ilteroi and @KungCheops for their contributions. Special thanks to Enginseer and Putmalk in particular for some incredible UI work, including many new icons.
This is also my largest rework of the Diplomacy AI in a long time, so I'm quite intrigued to see how the AI feels with these changes.
Fingers crossed for no gamebreaking bugs.
Changelog:
Link: https://github.com/LoneGazebo/Commu...download/Release-3.7.12/Vox.Populi.3.7.12.exe
Online as of 3:56 AM CST. Not savegame compatible.
New version inbound. I am exhausted so I'm just gonna drop the changelog and go. Many thanks to @Enginseer, @axatin, @seroperson, @Putmalk, @ilteroi and @KungCheops for their contributions. Special thanks to Enginseer and Putmalk in particular for some incredible UI work, including many new icons.
This is also my largest rework of the Diplomacy AI in a long time, so I'm quite intrigued to see how the AI feels with these changes.
Fingers crossed for no gamebreaking bugs.
Changelog:
Code:
General
- Performance improvements and code cleanups
- Added more defines to CoreDefines.xml and DiploApproachWeights.sql to tweak the AI's military strength evaluation logic
- Unhardcoded max. religions calculation
- Created GetQuestRewardModifier() and exposed it to LUA - returns a City-State's multiplier to yield bonuses from all quests (base = 100, PTP = 115, etc.)
Balance
AssignStartingPlots
- Blocked CS from spawning adjacent to ice
This means no more CS-exclusive 2-tile polar seas
However, this will also reduce the number of settling spaces available!
City-States
- Mutually exclusive buildings (including Wonders, for modmods) can no longer be selected for City-State building quests
- Added a long-requested feature: the ability to disable Influence gain from City-State quests (under Diplomatic Missions)
This allows humans to avoid anger from the AI over competing for the same City-States
While at the same time not canceling quests entirely, and other rewards are still granted
If a quest is received while Influence is disabled, and then Influence is enabled, the reward will still be the same as it would be normally
Influence gain from other sources (killing Barbarians, returning units, Self-Determination, etc.) is unaffected
This only prevents passive Influence gain from buildings, conquests, and so on
The AI will not currently use this
Not multiplayer compatible yet, so disabled in multiplayer (EDIT: Compatible as of 3.7.1!)
Communitas map updated to v2.6.2
- Increased Marsh spawn rate
Tile Purchases
- Tiles containing Barbarians can now be purchased, making them an exception to the rule that enemy combat units block tile purchases
But tiles containing embarked enemy units can no longer be purchased, making Barbs the only exception
Venice
- Can now invest in World and National Wonders in puppets, putting them into the Production queue (the city governor still won't choose to build Wonders on their own)
- Piazza San Marco now gains the +15 Culture from citizen births and +50 GAP from policies that the National Monument has
Deal AI
- Increased the price of third party war deals, now uses WarmongerHate flavor instead of War approach bias
Rationale: If the AI hates warmongers, they'll require more in order to go to war, and they'll also pay more to have others wage their battles (since they'll probably have $$$)
Diplomacy AI Rework
Diplo AI Options
- Vassals can no longer rebel if it would trigger a war blocked by the Passive Mode setting
- Added Diplo AI Option to disable coop war requests between AIs (set Disable Coop War Requests to 2)
- Changed the Diplo AI Option for Nuclear Gandhi: setting it to 2 now enables it even if Random Personalities is enabled
Unhardcoded it from Gandhi's UA, and now works in Community Patch only as well
- Disabled the Diplo AI Option allowing humans to trade their permanent items for the AI's temporary items (by default)
Due to complaints that the AI was always requesting Gold where GPT would be more appropriate
Can still be enabled in DiploAIOptions.sql
- Added "Limit Victory Pursuit Randomization" Diplo AI Option (not savegame compatible)
Disabled by default
Setting to 1 makes all AI players always choose the Default Primary Victory Pursuit from their XML hint
The override will not occur if the associated victory condition is disabled
If no hint, AI will still choose randomly
Random Personalities causes the AI to choose a random leader's Default Primary Victory Pursuit hint
Setting to 2 additionally prevents the AI from randomly choosing a Default Secondary Victory Pursuit unless hinted in XML
Setting to 3 makes the AI always pursue their Default Primary Victory Pursuit with no adaptation to the circumstances (unless disabled)
Domination Victory is considered valid even if disabled if it is possible for the AI to eliminate all other players
IMPORTANT: Options 1 and 2 only increase the likelihood of the AI going for that victory condition, while 3 guarantees it.
Diplo AI Leader Flavors
- Pacal War Bias: 6 -> 5
- Augustus CS Conquest Bias: 10 -> 12
- Genghis VictoryCompetitiveness: 7 -> 10
- Genghis CS Bully Bias: 10 -> 12
- AI now considers itself to be winning a war when their war score is 30 minus their DiploBalance flavor
- AI now considers itself to be losing a war when their war score is -20 minus their Meanness flavor
- Diplo AI flavor minimum/maximum values are now hardcoded to 1 and 10
Too much depends on these values being as they are for customization to work here
- The variability of durations for timed opinion modifiers based on flavors is now more compatible with customized durations
Previously the flavor's difference from 5 was added once for a < 50 turn duration, twice for 50, and thrice for > 50
Now it is modified by +/- 20% depending on the flavor's difference from 5, which works for any duration
Removed upward skew in the results from 1 being the minimum value; values below 5 decrease by 5% each, values above 5 increase by 4% each
Added support for inverting the effect of a flavor (high value = lower duration, vice versa)
Made it easier to swap out one flavor for another in the DLL
- More conditional (indefinite duration) modifiers now vary in value based on AI flavors
The change in value ranges from -50% with a flavor of 1 to +50% with a flavor of 10 (no upward skew here either)
Added support for inverting the effect of a flavor (high value = lower modifier, vice versa)
For the full list of affected modifiers, see the "Other Diplo AI Changes" section below
- Added a hidden feature to four flavors: when set to -12 in XML, they are treated as 12 (always 10) in the DLL, and...
Boldness: that particular AI will always pursue Domination Victory, as if Limit Victory Pursuit Randomization = 3
MinorCivCompetitiveness: same but for Diplomatic Victory (was DiploBalance prior to 3.7.3)
WonderCompetitiveness: same but for Culture Victory (was DoFWillingness prior to 3.7.3)
WarmongerHate: same but for Science Victory
Does not work for other flavors
Overridden by Limit Victory Pursuit Randomization IF that leader has a non-disabled Primary Victory Pursuit hint in XML
Enabling Random Personalities with a -12 value in XML makes it randomly selectable, and it still works in this case
Diplo AI Evaluations of Military Strength
- Rewrote associated functions to fix several nasty bugs
The biggest one being the AI not properly resetting the value added from Defensive Pacts when iterating to the next player
- Greatly improved AI's evaluation of particular ally versus enemy matchups during their strength estimates
AI will take advantage of its military alliances more often, both offensively and defensively
- AI now explicitly compares the enemy's attack power to its own defense power (including a % of allies' strength) to determine military strength
- AI now explicitly compares the enemy's defense power to its own attack power (including a % of allies' strength) to determine target value
Target value also factors in economic strength, this is unchanged
Thresholds for each scale point should now better reflect comparative strength
- Previously target value in particular was nebulous and essentially a copy of the military strength estimation
It is now a 7-point scale like military strength estimations (was 5), for additional granularity
- Better factoring in of allies' contributions to war state and peace willingness
- AI now factors Roman, Sweden, BNW Mongolia, BNW China, BNW Ethiopia UA bonuses properly when calculating military strength
Removed some of the recently added City-State war fervor from Rome AI to avoid being excessive
- AI's human strength perception bonus now applies more consistently, especially on higher difficulties
To elaborate, the AI adds "skill rating" for actions which increase war score and subtracts it for those which decrease war score
This makes the AI treat you more seriously if you run their units through a meat grinder, even if your military isn't as strong as theirs is numerically
Those below the global average would get a penalty capped at -50% to their strength perception, and those above would get a bonus capped at +100%
The problem was that if you were far enough below the AI in skill rating, you wouldn't get the base human strength perception bonus
Which made the AI not take humans as seriously as they should in a lot of scenarios
This has been changed so that humans' strength perception bonus will often apply if they haven't lost anything, even if they haven't conquered anything yet
The "buffer" where the human strength perception bonus falls off gradually as you incur losses has also been altered to apply more consistently
- Fixed an issue where the AI would double-count the military "skill rating" modifier to military strength when only one other civ is alive
- Fixed minor but annoying issue with military rating decay not being capped properly
- AIs now underestimate their opponents' military strength and target value by 3% per point of Boldness (30% with 10 Boldness)
AI also calculates and stores the raw value with no reduction, which it uses in some situations where extra caution is called for
Removed some "GetBoldness() > 6" type checks to determine if the AI would take certain modes of behavior, as these thresholds were sometimes arbitrary
- The AI now updates its estimates for military strength, economic strength, and defensive target value at the same time
Instead of doing so in separate functions (requiring more iterating across players)
- AI aggression towards the weakest civs, especially those lacking allies compared to the AI, has been increased
- On the flipside, AI will be more wary of enemy alliances than it was previously
- AIs' adjusted strength perceptions will also make the AI more willing to join its friends' wars
- Reduced AI cheating when factoring in war allies (AI can no longer view the enemy AI's opinions of players)
Other Diplo AI Changes
- AI early game aggression nerfed slightly, but the alliance-making changes should buff aggression indirectly
- Increased AI desire to dogpile-war conquerors
- AIs going for Science Victory are more likely to be friendly to nearby players with a strong military that are behind them in tech
- If only two players are left and AI is going for conquest, only 100 weight is added to WAR (was 1000) unless AI is close to winning
- Increased Montezuma's aggression towards early game competitors and (when not in a Golden Age) neighbors
- Increased Assyria's friendliness towards civs behind him in tech and aggression towards civs ahead of him in tech
- AI more likely to adopt the DECEPTIVE approach, particularly if allowed to shift from DECEPTIVE directly to WAR
Whether they're able to do this or not is based on personality and circumstances
High DenounceWillingness/DECEPTIVE bias civs are now more likely to be able to do this
Likelihood reduced slightly for mid-to-late game civs with conqueror UA/default victory pursuit
Blocked AI from being able to do this to players who have recently liberated one of their cities (most of the time)
- Removed some code from approach selection which looked at existing military alliances as it would now be double dipping
- Increased AI aggression when they are bankrupt and are paying a player a *lot* of GPT
- AI properly factors in proximity when scoring Defensive Pact choices and in some other calculations where it was bugged or misleading
- AIs who have conquered other civs are slightly more willing to tolerate multiple wars
- Increased AI desire to befriend and conquer City-States that have original major capitals in their possession
- Made Nuclear Gandhi attack City-States very rarely, he has bigger fish to fry
- Minor adjustments to peace and coop war desire logic
- AI no longer applies a warmongering penalty when it pays you to declare third party war, fixing that longstanding issue (city captures still incur penalties though)
- Removed mechanic where AI warmongering penalty towards their best friend/DP decayed at 3x the normal rate instead of 2x
This affected humans and AI alike, but on balance probably favored humans more
Removed because it only applies to AIs' penalties and skews data on how balanced (or not) warmongering penalties are
- AI approach selection modifiers from diplomatic alliances now scales based on the same flavors listed below
Except the AI's approach modifiers for having a DoF/Defensive Pact with them, which are unchanged
Added an approach penalty for being denounced by the AI's friends/allies
This penalty can also make the AI less GUARDED/AFRAID of the player, if a strong nearby friend has denounced them, based on that friend's estimated DiploBalance (3.7 to 3.7.2), DenounceWillingness (3.7.3+)
Especially in the early game!
- Adjusted some conditional (indefinite duration) opinion modifiers
Competing for World Wonders
Timer is no longer frozen when Wonder Spammer penalty active
Fierce competition: base value of -50 opinion (was -30)
Strong competition: base value of -35 opinion (was -20)
Weak competition: base value of -20 opinion (was -10)
Scaling is unchanged
Denouncing their friends
Extra penalty for denouncing the AI's "best friend": base value is now -20 opinion (was -15)
Extra penalty for denouncing the AI's "best DP" (if also a friend): base value is now -10 opinion (was -15)
If both penalties apply to the same player, they no longer stack, instead the highest value is applied
Penalty now scales based on AI's Loyalty flavor
Denouncing their enemies
Bonus is now applied for denouncing a leader they're at war with, not just one they denounced
Extra bonus for denouncing the AI's biggest competitor is now +20 opinion (was +15)
Extra bonus for denouncing the AI's biggest World Congress rival is now +20 opinion (was +15)
Bonus now scales based on AI's DenounceWillingness flavor
Being denounced by their friends/allies
Base penalty is now -15 opinion (was -10)
Extra penalty for being denounced by their teammates or best friends/DPs is now -15 opinion (was -10)
Check for which player has the better opinion now uses last update's cached score, not the cached rating
Bonus now scales inversely (negatively) based on AI's DiploBalance flavor
The more the AI is focused on their own sphere of influence, the less they get involved in others' fights
Making a Declaration of Friendship with them
Bonus now scales inversely (negatively) based on AI's DoFWillingness flavor
AIs more willing to be friends place less value on individual friends & more on forming blocs
Vice versa for AIs who want fewer but closer friends
Making a Declaration of Friendship with the same players
If both extra bonuses (best friend/DP) apply to the same player, they no longer stack, instead the highest value is applied
Bonus now scales based on AI's DoFWillingness flavor
Making a Declaration of Friendship / Defensive Pact with their enemies
Penalty now scales based on AI's Neediness flavor
Making a Defensive Pact with them
Base bonus is now +20 opinion (was +15)
Extra bonus for being the AI's "best DP" is now +20 opinion (was +15)
If AI has conquered another player, the bonus is now always increased by 50%
Otherwise, the bonus now scales based on AI's WarmongerHate flavor
Making a Defensive Pact with the same players
Base bonus for first DP is now +15 opinion (was +10)
Base bonus for subsequent DPs is now +10 opinion (was +5)
Extra bonus for making a DP with the AI's best friend is now +10 (was +5)
If both extra bonuses (best friend/DP) apply to the same player, they no longer stack, instead the highest value is applied
Bonus now scales based on AI's WarmongerHate flavor
"Your friends found reason to denounce you!"
Penalty now scales based on AI's DenounceWillingness flavor
Recent Diplomatic Actions
Removed all bonuses/penalties to this for denouncing someone
Penalty for rejecting a coop war request without warning the target: removed slight upward skew
Penalty for rejecting a coop war request and warning the target (with friends of the asker): removed skew, fixed bug
Vassals of the Same Master
Bonus buffed to +60 opinion (was +30) if same type of vassalage (voluntary/voluntary, capitulated/capitulated)
Tactical AI
- Fixed issue with invalid tactical moves
- Fixed issue with melee units dancing in place
- Tactical AI is now better at safeguarding civilians and embarked units
UI
- Enginseer: Added new FontIcons for Modmodders (MASSIVE THANKS TO PUTMALK FOR HELPING):
ICON_VP_SAPPED
ICON_VP_WRITER
ICON_VP_MUSICIAN
ICON_CSD_CIVIL_SERVANT
ICON_VP_SCIENTIST
ICON_VP_ENGINEER
ICON_VP_MERCHANT
ICON_VP_ARTIST
ICON_VP_AUTOMATION
(Gray?/Black? Specialist. Only used when you need to assign an Automation Citizen)
ICON_VP_BORDER_OBSTRUCTION
(TO-DO: This font icon exist so that maybe there's a way to detect the Great Wall and Ostrog in the future?)
ICON_VP_NOINFLUENCE
ICON_VP_GREATMUSIC
ICON_VP_GREATART
ICON_VP_GREATWRITING
ICON_VP_ARTIFACT
Added Artifact Icons for CultureOverview
Added Enable/Deny Quest Reward Influence UI
Mine Field UI Update
Airport City UI Update
Soft incompatibility warning for mods needing to update: CityBannerManager.lua, CityStateDiploPopup.lua, CityStateStatusHelper.lua, CityView.lua
- Lots of text improvements and clarifications, many of them from Enginseer
- City-States now display in their window tooltips how their personality and type affect yield rewards from quests
- Great Merchant trade mission now gives more detail about the scaling WLTKD bonus you'll receive
- AI now scores belief choices and recommends its top three picks for the player in the UI as trophy icons
Also added a debug option to show all the AI scoring, which may be useful for debugging
- Improved information on Historic Events in the Civilopedia
- "make siheyuan roof texture less annoying", sayeth ilteroi
- SoI/Open Doors now shown in non-EUI Diplo Panel
- Oversupply warning no longer says you cannot purchase units (because you can until you are 14 above it)
- Players may send a "stop converting" request to leaders that have not sent but have founded a foreign religion
- Fixed VictoryProgress display bug for Domination Victory in some cases
- Natural Wonders civilopedia entries now account for Adjacent Plot Yield boosts and Yield per Era
- Fixed Polder appearing twice in Civilopedia
- Unhardcoded special WarmongerHate descriptors for Aztecs, India, Venice
Other Bugfixes
- Fixed a major bug with the scaling of unit maintenance (thanks axatin!)
- Fixed several more desyncs in multiplayer (thank you seroperson!)
- Fixed Media Culture tenet not granting +25% Tourism to cities as advertised
- Fixed a bug where it was possible to work a different city's center tile with the Tradition range boost, causing several glitches
- Fixed a bug preventing submarines from attacking depending on the game's internal "area" borders
- Fixed "Route To Mode" not working properly for railroads
- Fixed Helicopter Gunships being affected by Mine Fields
- Fixed tile improvement progress UI bug
- Fixed Scrivener's Office having the wrong Production cost (now 125, was 150)
- Fixed AI getting an unintended extra 25 Gold, scaling with era, from clearing Barbarian encampments on any difficulty
- Fixed Embassy improvement not getting +2 Tourism from Futurism
- Fixed the spy missions to steal a GW / tech being canceled during mission preparation if nothing can be stolen
They should continue, and a Golden Age will now start if no tech/GW is available when the mission is completed
- Fixed resources located under captured city still remaining after losing the city
- Fixed supply free units from Conscription not spawning when already at supply limit (I think)
- Fixed the AI erroneously applying its WAR approach bias to its HOSTILE approach in several circumstances
- Fixed an issue with disabled event tooltips
- Fixed "GameEvent.IdeologyAdopted LUA callback is called before ideology is set as unlocked"
- Fixed an issue with Natural Wonders
- Fixed various Venice-related issues
- Fixed vassal rebellion being allowed if "Permanent Peace" or "No Changing War/Peace" game options are enabled (VERY edge case scenario)Link: https://github.com/LoneGazebo/Commu...download/Release-3.7.12/Vox.Populi.3.7.12.exe
Online as of 3:56 AM CST. Not savegame compatible.
Spoiler Newer 3.7 Versions :
Last edited:
