I considered it before, but I figured there were several major drawbacks (at least for the sort of campaign I'd like to make);
1/ The story isn't secret; and people rarely resist temptation
2/ Part of the attraction for campaign-makers - or at least for me - is the concept of ownership. That is, everything you make is yours until release - you can coddle and reshape all the little secrets and surprises right up until the final minutes before release. An OS campaign would remove that, IMO.
3/ Incohesive; eventually you need to have one person calling the shots or you end up with a muddy story compromised with too many ideas. And too many giant capships 
4/ Potentially disorganised; if it's 'open' to people dipping in, it's quite possible no-one will volunteer to do the more boring, sedate work. Who wants to model a sentry gun when they can make the GTD Murderousvengenacecoolthing?

1) I thought I'd have that temptation as an admin, but honestly? I didn't. Yes, I visit the internals of some projects from time to time, see if they have any cool screenies, but the storyline? I'd rather experience it playing it than reading about it.
2) Ownership of nothing is still nothing, unfortunately. If a campaign isn't getting made in the usual way, then there's no real drawback as far as that's concerned.
3) Perhaps I was unclear - indeed, I believe I was. I'm thinking of one person in charge (the Supreme Commander

), guiding a team of whoever wants to in hammering out the mission outlines according to his storyline. The actual work on missions, models, and whatnot would be the primary open-source venture. After all, even in the open-source browser Firefox, you didn't have people coding in scientific calculators (much to CP's disappointment, I'm sure) or hard disk defragmenting abilities. The goal was a browser. In this idea, the goul would be to bring a certain storyline to life in-game.
4) I don't think that will be as much of a problem as one might think. Those "boring" sentry guns can be a lot less intimidating (regardless of the modeler's experience level) from a time-perspective. Less commitment required to bring it to completion.
