For a community game, I think it is best to stay as stereotypical as possible. At least for the start. Small modifications can be made later. Firstly, you want to keep the unique ideas to your own game just so you can show off your creativity there, and secondly you want everyone to be on the same mental track. I doubt it'd take me less than a day to explain to someone what I intend Plagued Dead to look like. So for starters, maybe pick a game to mimic first in terms of how it plays, the feel of it, theme, etc.
Documentation I don't recommend for the programmer to do unless he/she "really" wants to. I personally hate it. I have grown to accept it, but documenting every thing I change can be such a huge pain. What you can do is put a little identifier comment next to stuff you change. Those who I have done work for have probably noticed these comments in their code, which look like:
Just throw one of those next to what you change, then if you want to document your changes, you can do that later. If you have a ton of changes, you can just give them a name and store the different names elsewhere like in a little text document:
'//!! Critical Hits
Stuff like crafting, you will want to wait until the project is decently finished. The less programming that has to be done at the start, the better, because it gets you a game out faster along with it keeps the interest of those who are participating. It is quite discouraging spending 10 hours on a project and seeing absolutely no visual change.
The EXP and Levels thing like I am doing for my game, that could be done easily enough, but it might be better to stay with something traditional, at least for now. That kind of system is more aimed for having a lot of stats, too, so you can really modify every detail of your character. It will also be easier to figure it out if was done more traditionally, though. But thats just me - ultimately, its your call. Server management and hosting I can handle at least for a while, preferrably when it is newly released to the public so I can catch the errors and fix them myself.
You did a good job listing off things that need to be done through-out the whole project, but I think it might be better to shrink it down a bit to what needs to be done now.
I'll be willing to help here and there, but don't expect a whole lot from me, at least right now, since I got vbGORE, Plagued Dead, and a new MMO engine I'm making from scratch all on my plate, and I "still" want to freakin' get time to learn C#.