A bit of history for the youngs. The custom of distributing preprints started out in high-energy physics. In the old days, pre-publication papers (literally on paper) would be sent out to university libraries by postal mail, often even prior to submission to journals, certainly before acceptance. Because this was expensive, the practice was mostly limited to national labs and other large institutional groups. The labs even had distinct letterhead for their preprints, and an internal numbering system for indexing. The libraries would then put them in a display rack once a week, and to update yourself on the latest, you would go into the library and sit at a table and read them. (They could not be checked out!) This was how it was when I started as a grad student. It was an informal and collegial system for communicating the latest science that flew under the radar of the journal system. In the early '90s, Paul Gisparg and Joanne Cohen (who is often left out of the story) set up a server to distribute preprints by email. You would receive a daily list of papers by email, and could request the TeX source via email commands, then compile it yourself locally. Being of a hacker-ish bent, I was one of the first in my department to embrace this wild new idea. Note that this was before the release of the first widely used web browser, NCSA Mosaic. The Wold-Wide-Web at the time was one stub site at CERN that you could find on Gopher. When Mosaic came out and the xxx.lanl.gov server went live with its Jolly Roger logo amd you could download preprints directly in Postscript format and print them on your department's expensive laser printer, the world changed forever. But it was never about papers already accepted for publication.