Also agree with grace. A simpler way to create and engage community (Google Groups, Posterous Groups, ...) may be enough for the early alpha stage.

About the software, I'd usually go with the "use existing software" approach. However, with the specs outlined (game mechanics and so on) there's also a big risk on ending creating a big patch "hacking" the existing software.

A third option in between would be to use some kind of framework as Django (or Ruby) with as much already done apps for the functionality as possible. For example, a comparison page for forum appz solutions on Django: http://djangopackages.com/grids/g/forums/