I've got Japan on my continent, currently medieval, and he's got a hard-on for Washington for some reason. Now, normally I wouldn't care, but they're the only nearby civ that's friendly and they're important trading partners for me. (Research agreement pending plus a key resource.) I've already paid off Nobunaga once to stop a war with America, but then he came to me requesting I join him in another war - which of course I didn't, but it means that yes, Virginia, he's going to start yet ANOTHER war (and probably destroy him), indirectly screwing up my economy. I tried making a defensive pact with America (and saving), no joy, he went to war anyway. (My army is stronger, but he evidently didn't care.) Now, I don't want to go to war with Japan either, same deal - research agreement, trade partner. Edit: reloaded and tried another tack... So, I've decided to try blocking his access to the American cities (they only have two) with my "neutral" troops in the area. (Pissy little scouts and bowmen, but they'll do for taking up a tile.) The question is - is the AI clever enough to see that I'm messing with their attack by stationing my troops on their attack routes? If so, welp, sorry, Washington - I'd rather you didn't die, but I'd rather not go to war with Japan pre-riflemen even more. If not, that seems to be the way to keep Washington alive long enough to buy Nobunaga off AGAIN, ie, after the 10-turn "no peace possible" period has expired. (Also keeping a scout on America's "cotton" resource tile to keep it from being razed.) TL;DR - if I block an AI's attack with my "neutral" units, is it clever enough to figure out I'm screwing with it?