arXiv is great, but represents a very small fraction of the total published work, is extremely field specific, and old papers are still very relevant. I find myself regularly relying upon citations from as early as the 1910's.
It's not even close to an alternative.