I was thinking that perhaps the ai should be required to hold gold in reserve (although this could be tied to civ personality), after the tech, currency, as a way of protecting themselves from wars they cant win.
I have stopped several enemy ais before they have captured vital cities by merely bribing them.
I often find myself wishing losing civs had faught harder before being vassaled...
The ai also runs into alot of issues where they rarely have money to trade, for resources, maps, techs...
The AI already does hold gold in reserve, and it depends on personality (since several version back). The AI actually has more money than it shows in the trade screen. AI leaders don't like to make all of their money available in trades because they knows that human players will trick them into buying useless techs all the time... But in any case, you may notice that, say, Mansa Musa will tend to have more gold in the bank than other civs - and you can often use that to your advantage by using spies to steal his gold.
The AI still needs to be better educated about when to not upgrade units.. Currently it will upgrade almost anything it can afford to; and even though the AI gets significant discounts for upgrades, this is still a massive waste of money in most cases. If this upgrade problem ever gets fixed, then I would expect the AI to reach their target gold stockpile levels more often, and so they'd more often have cash available for other stuff.
As for the cost of suing for peace.. that's something which could certainly use some adjustments. When I rewrote the for evaluating potential war targets, I intended to use that same rewritten AI to evaluate peace deals - but I just never got around to it. Part of the reason I never got around to it is that I also want the peace deal evaluation to take some additional stuff into account which is already in the current AI, and it isn't clear how the two things should be combined...
A couple question for Karadoc
1. Maybe overall tech rate could be slightly reduced (tech costs increased), based on map size? (if it is already that way, maybe it could be increased a bit).
1. It already is. This is one of the factors which comes into play with that team-size tech cost adjustment thing we talking about earlier in the thread. To me, there is no clear answer for how tech costs should scale with map size. Each civs effective beaker rate will be higher on larger maps, because the civs tend to have more cities and more trading partners, but how much higher is hard to say; because it will vary greatly from game to game.
The tech cost adjustment that's in place at the moment is basically just Firaxis' guess, and based on some of the other stuff that they've guessed, they probably didn't put a lot of thought into this; but nevertheless, I don't intend to change it without some hard evidence that it's wrong. ie. I'd like to see some data from many games on different map sizes showing that games on larger maps consistently reach the late-game eras sooner than on smaller maps. (The difference may also depend on the map script.. so this should be tested with something common; either
Not Too Big or Small;
Fractal; or
Continents.) It would probably be best if the data came from autoplay games, so that it doesn't depend on human differences. ...
This isn't something I intend to test in the near future.
By the way, here's an aside related to that previous discussion: