Should we adopt a common standard of citing scientific papers or should people decide on their own whether they just add a link or cite the paper like you would cite it in a scientific paper?
I don't think we need formal citations, as long as we have a permanent way of referencing particular sources.
Just about all journals use the Digital Object Identifier (DOI) System. I think it's perfect for here since it provides a permanent URL to specific resources. I think that's what we should use for linking to published papers, but I also don't think it needs to be enforced.
Since this is the web, I tend to prefer links. It's usually impractical for people to dig up articles that can't be linked to. This post is a good example of readable, useful linking. Note I'm not commenting on the quality of the references, just that it is useful when done in this fashion.