I love the idea of NuRide, although I've never used it myself. Anyone have some first-hand experience? It's a ridesharing system that hooks up drivers with passengers via the web, and it's running now in a few cities. The way they get past the axe-murderer problem is by having participants sign up via their employers or schools. Maybe the reasoning there is that if employees or students do go psycho, at least they'll be traceable?
Two things that would help would be to put it on mobile devices and get rid of the requirements for joining. I expect that when API's for mobile phone services come out, which someone told me should happen within a couple of years, an open system like this will be written that anyone will be able to use. This would mean way more people using it, which means way more rides offered-- and at some point it would reach a tipping point where people use the service casually, without planning ahead, figuring that they'll be able to get a ride back from wherever they are pretty easily. You could just rely on it the way people in some cities rely on being able to catch a cab.
If so, some custom would likely bubble up to make it worthwhile for the person giving the ride, probably some system for estimating gas and toll expenses. As the classic 70's dashboard sticker warns, "Ass, Gas, or Grass: Nobody Rides for Free." (NuRide rewards drivers with gift cards from participating retailers-- maybe they get money or tax breaks for promoting clean air.)
As for the axe-murderer problem, I think it's less of a liability and insurance issue if it's freeware tapping into a publicly hosted database, rather than a single company owning and running the system. And on the user side, I think there are enough people out there who would trust their own judgment whether or not to get in the car. But I suspect that it might find trust and acceptance faster if it started out only running on Blackberries.