i'm really glad that someone has taken this thread up. it seems like their are a lot of interesting scenarios that could play out with a re-imagining of nuclear weapons.
i read some book about nuclear war once. so i'm no expert, i don't know if the book was written by an expert, but i remember that during the cold war, some weirdly optimistic people thought that a tactical nuclear war would precede a total nuclear war. so like, the ruskies invade poland, NATO mobilizes some tank divisions and the conflict reaches a stalemate. so to break the deadlock, one side nukes a few tank divisions. it's not until later that a total war breaks out and we're into strategic nuclear strikes, like hitting cities and civilians.
so, the point that i'm trying to make, which has been touched upon by others, is that to build a "better" system to decide when the ai will launch involves mostly setting up some if-then arguements. consider
MartinHarper's post, #67. different actions, such as losing cities, may induce the ai to launch. maybe some ai wait to launch until you enter their territory. others might open with a nuclear strike. memorizing who does what might give the human player a slight edge, but knowing that monty is a maniac is general knowledge and yet it doesn't make starting next to him any less difficult.
Yak: You will note that the AI tells you if it has a nuke, doesn't it? So it had already said the threat "if you go to war with me, I will nuke you". You went to war with the AI regardless.
Dom Pedro II: True, but in many of these cases, it's in the AI's best interest to end the war as quickly as possible, so it couldn't hurt for them to give an ultimatum before they start shooting. The only reason why I don't think this would work though is because the human player doesn't get the diplo screen until the start of his turn. So unless the it was coded so that the AI's ICBMs fired out-of-turn upon refusal by the human player, the AI would have to wait until the next turn, which would give the human opportunity prep.
i think it would work out just fine, despite giving you a turn to decide what you're doing. here's how it would play out: ai says, "end this war or be destroyed." you can respond several ways 1) peacefully ending hostilities, 2) continuing conventional hostilities, 3) escalating to nuclear strikes. on the flip side, the ai could just start nuking you without offering an ultimatum, so it doesn't matter if you get a turn to prepare for a nuclear strike or not. i think that to make this feasible game-wise would be to code some ais, like gandhi, to offer an ultimatum, while the psychos wouldn't, then give each ai a threshold related to how many turns they wait.
i think since we're looking at nukes, we should take a pass at nuke defense. although this is outside the scope of the thread, changing the ai's approach to nukes might warrant altering some of the other nuke related game rules. i think the SDI should be nerfed (also, it's technologically impossible in RL) and that it should be bumped down to 50% protection from ICBM's and 25% for tactical. however, and maybe it already does, but it would be effective globally in order to offer some protect to your conventional troops who are far afield. on the flip-side, bomb shelters need to be buffed. so troops fortified in a city could have their damage capped at 50%, and populations loss capped at 8, maybe. basically, this would force nukes into being a more defensive weapon. if you could code the ai to really fear them, they could be used realistically as a DOW deterrent.
one last thought is that espionage spending could give you some indication of how many troops and nukes the the ai has total in the F5 screen, without revealing their positions. stop me if it already does this. i mean, eventually you can get city view, but i'm talking about something before that.
