...SWORD is not a harvesting protocol; it is a deposit protocol. The party that initiates a SWORD deposit is the party with the material in hand. SWORD offers no way for a repository that wants material from another repository to request it, much less do so in an “automatic” fashion....
(If PubMedCentral or arXiv or RePEc or SSRN were to implement SWORD, it would be a gigantic step forward. Please, will someone in the know suggest it to them?)...
Paul Ginsparg replied by email (posted with permission):
arXiv has implemented the SWORD protocol for ingests (in collaboration with Microsoft) -- it's currently used for the direct upload from the Microsoft Word plug-in and is also being tested for upload from CS conference software (which manages only up to the point of the conference, so needed some way to pipe output to archival holder of proceedings).
In principle, IRs would be able to "push" content to arXiv if they were set up to export via SWORD, and were able to map to the arXiv metadata format via atom extension elements (vanilla SWORD doesn't mention metadata mapping). Though according to the Xia article the holdings in some of those IRs may currently be problematic (spotty metadata, broken links to full text). Current self-archiving authors could perhaps be spared some duplicate effort via this route, but they'd likely want to see some natural functional advantage to it, rather than viewing it as a requirement.
Anyway we agree this is all in the noise, and the real issue is the current 85% non-self-archivers. Ultimately they shouldn't need to be required to deposit open access content any more than people needed to be forced to deposit to youtube ...
Update. Here's a follow-up from arXiv's Simeon Warner (posted with permission):
I think that SWORD can significantly reduce duplicate effort. I think the most logical direction is for IRs to have the facility to assist with push to arXiv. Of course there is some arXiv specific stuff that is likely not normally present in the IR so there will be some additional effort, however quite a bit could be reduced to the "push to arXiv" button. This way one can use local effort/expertise (perhaps even with advantage of local language) to help with submission to centralized facilities such as arXiv.
The open access movement:
Putting peer-reviewed scientific and scholarly literature
on the internet. Making it available free of charge and
free of most copyright and licensing restrictions.
Removing the barriers to serious research.
I recommend the OA tracking project (OATP) as the best way to stay on top of new OA developments. You can read the OATP feed on a blog-like web page or subscribe to it by RSS, email, or Twitter. You can also help build the feed by tagging new developments you encounter.
