Tweet
I've observed that the ai handling of peace clauses defies imagination.
Here's an example, on Monarch, Small map, epic. 5 civs. I have one continent of my own, the Romans another, and the biggest continent is split between the Incas (many cities), the Indians (4 cities, +2 on another small continent), and the Japanese (2 cities + 1 on an island).
The Indians ask me to declare war on Tokugawa. Since the Japs are the last on the power graph, are never worth trading with, I say ok. I send some troops, pillage several towns, disrupt their economy, and, although I lose my troops, I've totally destroyed their economy. So I want to sue for peace and send my ships elsewhere. Tokugawa asks for liberalism. Errr??? Sure, He lost maybe 4 land units and destroyed a dozen, but does that mean he has an edge? He lost all his ships, has no way to invade me... I send troops near a size 1 city he founded on an island near me. There's one longbowman inside. I have 2 grenadiers and one catapult. I sue for peace: he still demands liberalism.
So I take the city, and now he's willing to give me world map and all his gold (only 90) for a peace treaty.
This is obviously stupid. It's so blatantly unnatural that I'll probbaly skip that game and start a new one because it doesn't make any sense.
The ai should evaluate the current state of the war:
-It's unable to launch an offensive, my ships are blocking it and making all his sea plots unusable, I have overwhelming forces near one of his cities.
Rather than based on history or whatever silliness it uses:
-It managed to destroy all the troops I landed on his continent (but these troops did what they were there for: They destroyed all his towns and most roads - thus it was not a victory for the japanese: I reached my goal).
Even if it did, it should really compute the value of the damage done and being delt to its economy: 2 fishing boats destroyed = -12 food per turn in the capital. 4 towns destroyed = -20 gpt for tens of turns, etc. Right now it doesn't look like it takes these parameters in consideration.
Does someone understand what's the algorithm behind the peace negotiations? I hardly ever manage to get a tech out of an ai (typically the ai will have something like Divine Right, which I never research), even when it's only got one city left.
The way it is coded, it looks very worth trying the following 'tactics': When at war with a strong opponent, if they have an isolated city (island, small continent...) - try to take this city. Just defend for 10 turns. When they want to talk to you, they'll be very likely to make peace because you took one of their cities, even though they could probably take many from you in one turn or two. This will be particularly true if you managed to destroy some of their units using guerilla warfare.
Here's an example, on Monarch, Small map, epic. 5 civs. I have one continent of my own, the Romans another, and the biggest continent is split between the Incas (many cities), the Indians (4 cities, +2 on another small continent), and the Japanese (2 cities + 1 on an island).
The Indians ask me to declare war on Tokugawa. Since the Japs are the last on the power graph, are never worth trading with, I say ok. I send some troops, pillage several towns, disrupt their economy, and, although I lose my troops, I've totally destroyed their economy. So I want to sue for peace and send my ships elsewhere. Tokugawa asks for liberalism. Errr??? Sure, He lost maybe 4 land units and destroyed a dozen, but does that mean he has an edge? He lost all his ships, has no way to invade me... I send troops near a size 1 city he founded on an island near me. There's one longbowman inside. I have 2 grenadiers and one catapult. I sue for peace: he still demands liberalism.
So I take the city, and now he's willing to give me world map and all his gold (only 90) for a peace treaty.
This is obviously stupid. It's so blatantly unnatural that I'll probbaly skip that game and start a new one because it doesn't make any sense.
The ai should evaluate the current state of the war:
-It's unable to launch an offensive, my ships are blocking it and making all his sea plots unusable, I have overwhelming forces near one of his cities.
Rather than based on history or whatever silliness it uses:
-It managed to destroy all the troops I landed on his continent (but these troops did what they were there for: They destroyed all his towns and most roads - thus it was not a victory for the japanese: I reached my goal).
Even if it did, it should really compute the value of the damage done and being delt to its economy: 2 fishing boats destroyed = -12 food per turn in the capital. 4 towns destroyed = -20 gpt for tens of turns, etc. Right now it doesn't look like it takes these parameters in consideration.
Does someone understand what's the algorithm behind the peace negotiations? I hardly ever manage to get a tech out of an ai (typically the ai will have something like Divine Right, which I never research), even when it's only got one city left.
The way it is coded, it looks very worth trying the following 'tactics': When at war with a strong opponent, if they have an isolated city (island, small continent...) - try to take this city. Just defend for 10 turns. When they want to talk to you, they'll be very likely to make peace because you took one of their cities, even though they could probably take many from you in one turn or two. This will be particularly true if you managed to destroy some of their units using guerilla warfare.
Comment