There is another thread of excellent comments on this post started on Blog Around the Clock by Bora Zivkovic 'Coturnix'. Some common themes with those comments here.

In particular I strongly support the notion of the need for an ontology of roles. Without it you get two general problems: (1) too little resolution of roles leading to the current situation where everyone seems to contribute to just about everything. For example, see the contribution statement of this article in JAMA . Or (2) the roles lack consistent meaning, leading to more confusion and little value.

The question is how to get this started? Clearly some visionary editors and directors at BMJ [article], JAMA [article], PLoS [article], PNAS and other journals have recognized this problem for a decade. The challenges is how to agree on principles and propagate it to journals, authors, employers, tenure committees and such.

One possibility is to set up a shadow system, perhaps sponsored by a few of the well-respected journals, that would provide a contribution ontology and a database where authors could begin to add their explicit contributions. Call is the IScDB (Internet Scientist Database). (Someone better register that web address before it's taken!) All stakeholders can participate in the ontology building. And as it catches on, more and more journals can simply latch on to it and ultimately make it a publication requirement.