chronic lack of MizarDocumentation?, nobody wants to spend time on it, it gets outdated very quickly, new users get stressed and deterred; the incremental "users for users" model could make a difference
"circular" discussions about various features, caused by the lack of documentation ("Yes we discussed this 10 years ago, then forgot about it")
unclear feature/project? priorities, caused by low and undocumented knowledge about their interdependence
Cons
as in any collaboration, the risk that nobody will contribute
additionaly, the PublishOrPerish? pressure in science, forcing people to prefer their own publications (cf. the QedManifesto? Bourbaki-style utopia )
various MizarWebs? and MizarMailLists? exist already, their relation to Mizar Twiki is not clear (depends on authors/users)
authorities are established, no one will want to edit AndrzejTrybulec?'s comments
Implementation
takes most inspiration from the TWiki:Codev web, go there for additioal implementation ideas