One of the things we're doing is some scripting where things we like are concerned. One of the main things is seasonal changes, but the one I wanted to just outline was the Garrison script we're using. Most garrison scripts are there so that you attack an AI faction, and it produces a large response to this by adding garrison units to that place or as a response to your actions/success. It's accepted widely, but I don't like it, as I played a particular game and attacked early on, taking a major target quite easily. The next turn I got splatted there by a massively superior army, which essentially had the power to end my game, because I couldn't field troops that good myself, and they ended up annihilating me. They rolled up 3 settlements and I had to just give up.
I also liked Empire's thing of sprogging a few awful defenders.
SO, we decided to have a garrison script that means a slightly random amount of 'citizen-levies' appear to defend against a settlement attack; it's based on size, but with a little random element. They're not awful, and potentially can hold a city for you. SO, you attack, they have their base garrison troops, and you can add on a few units of these guys. They depend on faction, for weapons and the rest, but they will all disperse after any assault. If the city falls, they're dead, if it holds they become normal citizens again.
This is just one of the little examples of our work. There's nothing that's ridiculous, and we have a stable enough build, so it reduces the amount of chance for errors by limiting scripting. Everything we choose is for good reasons
We want to make sieges pretty cool! Very cool, as they're one of the main game parts. Hopefully the days of walking to a wall and it falling over for you are gone here. Not to the other extreme, but just a bit more uncertain.