2009/7/17 Thomas Dalton thomas.dalton@gmail.com:
Not really. For the kind of things you get on arXiv "reliable" is usually taken to mean "peer-reviewed", so you would cite the journal the paper was published in. You might link to the arXiv version for convenience, but you would be citing the journal, not arXiv.
"For convenience" is a good way of putting it. Some arXiv-published work is of minor note (e.g. Garrett Lisi's e8 Theory of Everything was in the news after its release), but being in arXiv is not in itself anything special whatsoever. It's all but self-published.
(Which is why viXra is such thrilling news. It's like Hell wasn't deep enough, so they're digging tenth and eleventh circles below the basement.)
- d.