Since this appears to be a topic that is really actually still in the 'how do we' phase, it's probably one of the more constructive we can engage in.
I've played a couple of games with some forms of sieging...a MUD (actofwar) which was less sieging and more capture the city, DAOC, altho the keep battles were more akin to capture the flag than anything else, and Shadowbane. I have not played DF's version of it, nor MO's...those games pitfalls kept me from playing them in general.
A couple key components that i see to a successful sieging system.
1. Difficulty. Conducting a successful siege should be hard. And expensive. Actually really hard and expensive.
2. Advantage. All things being equal, the defender should win. I real world tactics your rarely try to take an objective with less than 3:1 odds. All things being equal, a city should be able to capeably be able to defend itself against ~2:1 odds. If the defenders are more skilled, or the city exceptionally designed, this potential should get much higher. I have no problem with impenetrapeable fortresses, so long as they are impenetrable by player design and effort, not by code.
3. Fatigue. Winning or losing a siege should not come solely down to who can be up until 3am on a tue night (been there)...much less 3am on mon, tue and wed (been there too). Sieging should be a purposeful event that takes time to initiate, and there should be an interval between attempts.
4. Invulnerability. I am not a fan of 5am wall bashing. Assets (excepting perhaps gates??) should be invulnerable outside of a specific warring phase of a seige. i mention the gates, because being able to raid cities with purpose to raid vs. destroy would be nice....not everything needs to be for pinks. No one should have their city torn down at 4am when no one was able to be on to defend it.
5. Objective. When do you win? The easy answer is the totem. control the totem, you win. how do you control it? stand near for a period of time (like WAR), tear it down (like sb), maybe make it spawn a protector whom you have to defeat (ie. DAoC)? lots of different variables here.
6. Vulnerability. IMO everyone should be vulnerable...evil should be much more vulnerable to siege, but be able to siege at a lower cost, good should be the attacker meat a much higher level of [whatever - i made a case for thise previously] to induce vulnerability, and in turn have a much higher threshhold of [stuff] to obtain to initiate a siege. Example two values, attacker, defender. NUMBERS ARE ARBITRARY AND JUST FOR ILLUSTRATION: we'll say evil has a defense value of 10, and an attack value of 30. Good has a defense value of 50 and an attack value of 100. For evil to attack evil it needs to meet a value of 40. for evil to attack good, it needs 80. For good to attack evil it needs 110. For good to attack good it needs 150. of something along those lines.
7. Mechanics. Generally. Attacker places a seige totem within a certain distance of the target tribal area. Placement of the totem allows the crafting of seige weapons within the seige totem radius...for both offensive and defensive sides. A window of time exists for both sides to build (48-72 hrs?) During this time both sides camps are vulnerable to attack...so yes the attacker will be building a siege camp which they have to defend. Attackers would be able to tear down the defender gates and potential raid equipment building in the defendign city. The defenders sets the window for the seige to occur (48-96 hours after placement of siege totem. Once this ticker expires the war window begins, and ends when either the attackers capture the town totem, or the defenders capture the siege totem.
that should be enough to spur some discussion, maybe.