(7-55) Handicap from Great People Now Provide Yields Based on GP Type

[Legen]

Spoiler Current AI handicaps at Emperor regarding Great People, from DifficultyChanges.sql :

SQL:
-- Yield types given for each type of difficulty bonus trigger
INSERT INTO TriggerYields
  (HistoricEventType, YieldType)
VALUES

  ('HISTORIC_EVENT_GREAT_PERSON', 'YIELD_FOOD'),
  ('HISTORIC_EVENT_GREAT_PERSON', 'YIELD_GOLD'),

SQL:
-- Base amounts by Difficulty Level and Era
INSERT INTO DifficultyBonusAmounts
  (HandicapType, EraType, Amount)
VALUES

  -- EMPEROR
  ('HANDICAP_EMPEROR', 'ERA_ANCIENT', 72),
  ('HANDICAP_EMPEROR', 'ERA_CLASSICAL', 72),
  ('HANDICAP_EMPEROR', 'ERA_MEDIEVAL', 139),
  ('HANDICAP_EMPEROR', 'ERA_RENAISSANCE', 304),
  ('HANDICAP_EMPEROR', 'ERA_INDUSTRIAL', 359),
  ('HANDICAP_EMPEROR', 'ERA_MODERN', 512),
  ('HANDICAP_EMPEROR', 'ERA_POSTMODERN', 695),
  ('HANDICAP_EMPEROR', 'ERA_FUTURE', 906),

Proposal

• AI handicaps from generating Great People now provide yields based on which type of Great Person was generated.
  • Suggested yields and multipliers for this session, by GP type:
    • Great Writer/Artist/Musician: 1x Food and 0.25x Culture.
    • Great Scientist: 1x Food and 0.25x Science.
    • Great Merchant: 1x Food and 1x Gold.
    • Great Engineer: 1x Food and 0.5x Production.
    • Great Diplomat: 1x Food, 0.5x Gold, 0.1x Culture and 0.1x Science.
    • Great General/Admiral: 1x Food, 0.25x Production and 0.125x Science.
    • Great Prophet: 1x Food, 0.25x Culture.

Rationale

At the moment, Great People are a major source of Gold for the AI, and does so without differentiation of GP type. This alters how both civilizations and policies perform, as this Gold influx may cover their weaknesses, or interact with their strengths in unintended ways. Potential cases include Austria having way more Gold than intended due to their Marriage mechanic accelerating how often this handicap triggers (and in turn, how often they can afford marriages...), or Industry's investment efficiency working way better than it should (AI tests pointed to Industry having an unusually high win rate compared to Rationalism and Imperialism).

This proposal aims to align the handicap yields to the GP type, so that the handicap doesn't create a mismatch between the focus of civilizations and policies, and what they get in return. In addition, differentiating between GP type allows another way to balance the handicaps based on how strong each type is at any given patch. This is also a potential new way to balance warmonger AI performance.

The initial suggestion is based on the yields their respective specialists provide, with the modifiers being based around the usual evaluation that > > , and with the Civil Servant's current base yields (211) being roughly 0.5x merchant, 0.33x writer and 0.33x scientist. For military GPs, the yield choice was based on the base yields on the Citadel (11), adjusting them to be half Great Engineer and half Great Scientist for handicap purposes.

Great Prophets are treated as a cultural person for handicap purposes, as Culture is one of the base yields they can provide with Holy Sites (453).

MAGI: Complex proposal, DLL and database changes

 

[rkkn]

wasn't production removed from the handicap instant yields?

 

[pineappledan]

Just putting this here. This is some analysis that @L. Vern has been working on.

The highest correlation of any AI personality, approach, or flavour with winrate is for Great People. Essentially, the higher an AI prioritizes generating great people the more they win.

 

[Rekk]

Just putting this here. This is some analysis that @L. Vern has been working on.
View attachment 688410
The highest correlation of any AI personality, approach, or flavour with winrate is for Great People. Essentially, the higher an AI prioritizes generating great people the more they win.

Higher than even Personality: Diplomat at 0.529, Flavor:Offense at -0.512, Pursuit: Domination at -0.532, or Approach:War at -0.47?

 

[pineappledan]

Ah, missed those.

 

[Recursive]

I guess I can sponsor this.