By supporting Benedikt Mandl's proposal in-house, rather than being persnickity and forcing him to go outside for support, we guarantee such things as long-term software and content compatibility, etc.
This is by far the best argument you make. But from a technical standpoint, do we have the resources to give the necessary support? Would we would need an extra machine to make this happen?
As long as it's kept an unofficial project at first, and the developers aren't forced to spend too much of their time helping set up the Wikispecies project, I don't see too much potential harm.
Anthony