One of Richard Durbin's points which I think is particularly important and bears repeating, is that Pubmed (Medline) is a superior search tool. Although, in my opinion, OAIster is an excellent search tool, and distributed archiving a needed approach, when it comes to searching, no general tool can match a searching & indexing tool that is developed to meet the particular needs of a discipline. A search that begins with Pubmed and leads the individual to the fulltext provided through OA - regardless of where the article is archived - is the best means of connecting the user as directly as possible with exactly the information they need, in the medical arena.
If the articles are housed in a central server, then ideally they would also be able to be searched via OAIster as well - that way, users who are looking for other kinds of information besides the strictly medicine-based, will find what they need as well. May I also suggest that central vs. distributed archiving, with OA, is not an either-or proposition? An OA article housed at Pubmed can be easily included in an institutional archive and the author's own website as well. Given that the most basic of technology issues regarding the archiving and preservation of material in electronic format have yet to be worked out, the safest approach, and the one I would recommend, is all of the above (central plus institutional plus author's own website). This would fit with the LOCKSS (lots of copies keep stuff safe) principle. It also seems to me that there is no reason why there needs to only one approach to OA. One approach might be more suitable for one discipline or sub-discipline than another. For example, if there is any group where the tendency to publish is relatively small because that particular discipline does not place quite the same emphasis on publish-or-perish as other disciplines, then perhaps publishing could have lower costs due to lower submission rates leading to lower rejection rates. Physics seems to do doing well with preprints, whereas in other areas it might be more important to ensure that readers looked at the corrected postprint. As long as the results are OA, the details of where and how things are published don't really matter, do they? Therefore, I would second Richard's suggestion that those who advocate for OA should be unanimous in our support for the NIH proposal. cheers, Heather Morrison
