In a posting to the MEDLIB-L discussion list, Lisa Blackwell raises the question of what to do about the REP (Rapid Electronic Publication) papers in Molecular Endocrinology. She wonders if we (i.e., librarians) should advise researchers to wait until the final manuscript is published before citing it, and whether or not it will become "a legitimate practice to fill ILL requests with author manuscripts."
This is just the tip of the iceberg. I've been working with a NISO group on "the Versioning Problem" for several months. Our charge has been simply (!) to devise standard language for describing and assigning metadata to the various versions of an article. We've had a series of bi-weekly conference calls and the discussions have been fascinating. Once you start to really dig into it, the number of potential versions and versions of versions is mind-boggling. It's not driven only by the early publication experiments of journals like Molecular Endocrinology -- and there are a lot of journals that are beginning to do this -- but by the NIH Public Access Policy, the move to establish institutional and subject-based repositories, and the increasing practice of authors posting versions of their articles on their personal/professional websites.
As librarians, we have to accept the fact that all of these versions count as "publications" -- that is, they are out there and are being read and used and, yes, cited, whether or not we consider them to be "the definitive version". In response to Lisa's specific questions, I would probably not advise someone to wait, although I would remind them that the final published version may vary in some substantial ways from the author's manuscript version and they may want to take that into consideration (it is also the case that sometimes the "final" version will never appear, but that's another can of worms that I won't get into here). What is critical is that they be sure to cite the version that they actually used. For ILL purposes we need to be sure that we are supplying what is asked for -- in other words, if a request is for the author's manuscript, that's what should be supplied. The author's manuscript should not be sent in lieu of the final published version, but neither should that version be sent if what one really wants is the author's manuscript.
Obviously, this complicates the lives of librarians tremendously -- but welcome to the world that we live in. When I spoke to the Elsevier managers last week, one of the things that I touched on was the need to develop more comprehensive search systems that will help us find our way through this increasingly complicated information space. There are a lot of intriguing and useful attempts being made, but we are only in the very early stages of what will be a long evolution of figuring out how to deal with the digital scholarly world.
The days when all we needed to worry about was tracking individual articles that were neatly and uniquely identified by volume and issue numbers are over. There is no turning back.