I think the lack of response here is a clue. Perhaps you've already considered it dead and moved on, but I guess here's one for the search results.
In general I've gotten very pessimistic. Commitment is the key problem and it's in sparser supply than ever. It's just hard to get people to care, and then to care long enough, and then to humor the advanced logistics that are going to hurt interest further if any of it is on the frontend, except bottling it up on the backend is a key cause of burnout. I think one literally can't 'just make' a persistent world no matter how much effort they put into the setup. It needs to be great circumstances with momentum. I'd be interested in trying one out, but I think the most likely issue above is going to hit no matter how much feedback you get, how much interest is expressed, etc. But I'll throw thoughts at some of the hypotheticals.
1. Staff are there to do things cleanly. Be efficient, but be accountable. I think there should be a divide between what people are pinged for and stuff that is available if people are interested. Keep anything serious exposed, but don't air the laundry - especially as far as discipline, which should be handled efficiently as well. Personally I have the 'draw the line, give a chance' approach in most things. Tell them something is up, tell them what it is, be open to clarification and keep it between you without making it a parade. If you must, act. Be open about why but don't let it linger. My take anyway.
2. Simple to get into, complex in detail. Don't make people read a million threads to start, but perhaps lure in their interest and then allow them to dig into things and get that 'oh that's cool I didn't even realize' rush once in a while.
3. Curate it. Have a center vision, allow expansion and approve propositions for - but be mindful of its impact on the setting and don't let people add willy-nilly.
4. Personally, no. I don't think there's much chance for something huge to work unless you kinda defeat the point and limit people to sections at a time in their own bubble-areas. This site's ill fated PW is a pretty good example. Better off smaller locations with more chance to overlap. As much as flexibility is great I've come to think that if there's a chance, it'll be from focused GMs standing their ground and not letting things spread too far. Build the core, don't just have a roleplay site with a vaguely connected idea of lore.
5. Haven't been on many dedicated ones, but I like decking out profiles for characters.
6. Bureaucracy in general. The backend should be pretty streamlined. You could blend roleplay into administrative things, but be prepared for that to cause issues. It can cause the right intrigue too, but be careful.