Hi. Some questions/feedback on ideas.
War desire reduction from other players' Defensive Pacts is now more nuanced
- Factors in proximity
Is this also factoring in proximity of DP partner to attacked nation? For instance, if Civ1 wants to DoW Civ2 with a DP with Civ3. Civ1 and Civ2 are "close", however Civ1 and Civ3 are "far", BUT Civ2 and Civ3 are "neighbors", such that Civ3 will easily be able to channel units into Civ2's territory, is that taken into account? Is that relevant in how the AI will act also (I'm not sure about how the AI calculates to involve itself in DP wars where it's not directly invaded or near the Civ1 in this instance).
Also a side note on the world wonder penalty (which I agree is a PITA as it was in the March version): I feel making it a comparison to the next highest wonder builder is going to lead to some "unpleasant" scenarios where that 2nd place guy loses a city and suddenly forces the player to get the penalty despite not building any new wonders. I would suggest an equation rather than strict cutoffs for penalties:
y=1.038462-1.538462/(1+(x/1.259317)^{3.169925}) //throw that in here to see curve
https://www.desmos.com/calculator
Where X =( [civ#wonders] / [#totalwondersbuilt] ) / ( 1 / [#alivecivs] )
gives a nice curve that gives a bonus to relations of 50% if you have 0 wonders, a 0 bonus when your #wonders is the same as your equal share per civ, and rises fast after that up to 100% negative relation modifier at 4x your "fair share" (so if there's 16 wonders built and 8 civs, if you've built 8 wonders you get ~100% neg relations mod).
Explanation:
1. I like whenever there's a negative relations modifier, there should be a positive one (albeit lesser perhaps). Essentially, AIs are using wonder distribution to choose DoW targets, and being a "wonder lite" civ should be helpful to relations likewise.
2. Polynomial curves instead of clear switches are preferable always imo. Perhaps not from a performance standpoint, but certainly from a general humans (and AI) relations standpoint we generally respond proportionally rather than on/off.
3. This curve basically maxes out around 4x your fair share, and rises quickly then sort of plateaus, which makes sense as a "a lot of wonders" and "a super lot of wonders" should be less of a difference in relations than "fair share" and "a lot".
Quick question: is all of this diplo/AI changes in the .dll or does it require the full VP?