What I meant is that the A.I. needs to have direct line-of-sight (i.e. a unit or agent standing near to it) on your fort before it will consider attacking it. Sorry for being unclear. I am pretty sure the A.I. sees the same campaign map as we do; I just think it ignores forts unless it has visibility over them. This is not entirely stupid, since forts disappear when left unoccupied.

I am speculating on how the A.I. uses armies, but I suppose it assembles a task-force with a mission in mind. If this mission is capturing a city, the army will presumably ignore other targets, such as forts. If this is correct (and that's a big if); when a fort blocks the armies path towards the target city, the A.I. will stop, reassess its route, conclude it is too long, and give up. I am pretty sure the A.I. is not so complex that it can decide to destroy the fort as a step on the way to taking a city.

Of course, once the A.I. has given up on reaching the city, it might well decide to have the army attack the fort. But I think the A.I. won't declare war for the sake of attacking a fort. For the sake of capturing a city, yes, but not a fort. And, as I wrote, I don't think the A.I. is complex enough to see taking a fort as a means towards conquering a city.


ETA: I just realised I am contradicting myself here. If the A.I. is unwilling to declare war by attacking a fort, then that would explain why a blocking fort is so effective at preventing attacks. The LOS-explanation is not required, and, going by ayekides' experiences, not sufficient to explain this.

Reply