As a journalist, I receive a lot of press releases that cite “forthcoming” papers. Depending on the publisher one can usually find the paper in a pre-press state on their website. However, it’s often the case that the DOI does not go live at the same time as the embargo expires on the press release, and so I might legitimately publish an article about the research I cannot use the DOI as the reference and must use the direct URL for the paper. Unfortunately, some publishers then move the paper when the paper publishes, so the link I used ends up broken.
Moreover, this cannot be useful for authors themselves in that a paper that does not make the grade at the International Journal of Good Stuff and ends up being resubmitted to the Parochial Bulletin of Not So Good Stuff will gain a different identification code along the way.
Will Griffiths on ChemSpider was recently discussing the possibility of an OpenURL system. I think we could go one step further.
A simple standardized way of generating a unique identifier for each and every paper that would be transportable between different phases of the publication process from submission to acceptance and publication, or rejection and resubmission elsewhere, would be a much better way of registering papers. The identifier would be created at the point when the final draft is ready to be mailed to the first editorial office in the chain, perhaps based on timestamp, lead author initials, and standard institution abbreviation. It could be the scientific literary equivalent of an InChIkey for each research paper.
There would have to be a standardized validation system, so that authors were sure to be using the right system, but that could be established relatively painlessly through the big institutions, be networked and have cross-checking to avoid duplicates. And, of course, be open source, open access.
The possibilities are endless, PaperID would create an electronic paper trail from author through preprint, in press, to online, and final publication. It might even be back-extended into the area of Open Notebook Science and equally usefully into archival, review, and cross-referencing.
DOI is useful most of the time OpenURL sounds intriguing, but PaperID could be revolutionary.