Obviously the diplo modifers are there to feed back data to the human and the diplo rep penalties are there to keep human players somewhat honest. The AI itself doesn't have to protect itself against bold faced lying.
I've long proposed a mechanism to feed back the same diplo modifers AI uses back to the game (but for the human) but with a system like that, you invariably run into cheating and lying and psycopathic human behaviors
So you'll need a scoring mechanism to go with that to track honesty. So that when we said we are friendly with X, did you really mean it etc etc but because such a system can only rate you after the fact, it can never be accurate in feeding back the right data to the AI. So the current reputation penalties for 'breaking your word' is as good a penalty as it can get.
We're just making sure they are working properly and don't punish players for liberation.
I've long proposed a mechanism to feed back the same diplo modifers AI uses back to the game (but for the human) but with a system like that, you invariably run into cheating and lying and psycopathic human behaviors
So you'll need a scoring mechanism to go with that to track honesty. So that when we said we are friendly with X, did you really mean it etc etc but because such a system can only rate you after the fact, it can never be accurate in feeding back the right data to the AI. So the current reputation penalties for 'breaking your word' is as good a penalty as it can get.
We're just making sure they are working properly and don't punish players for liberation.