Steve said the only way is using OAi-compliance by the author to selfarchive his documents before and through refereeing.
The word only is too much of a load. In Physics (and Mathematics) since a long time authors can selfarchive their documents, without having to install any software or learn about OAi. They are automatically included into the OAi scheme by the OAi compliant service providers by using PhysDoc (or Math-Net) as gateways who take care of their document being included. How does it work: 1. author puts his document on his own server. 2. author fills webform http://physnet.uni-oldenburg.de/services/mmm/ and adds or attaches the resulting (for him ununderstandable metadata infested 'shadow file' to his document. [for an example of a shadow file see my own documents, back to 1968, see http://www.smallsystems.de/publications/metadocs/ebs.kernspaltung.html or all of http://www.physik.uni-oldenburg.de/Docs/THEO3/information/publications/] 3. The document is found by PhysDoc-OAi compliant service provider, see http://physnet.uni-oldenburg.de/oai/query.php The PhysDoc crew takes care to track all (for shure to come) changes of the OAi compliance regulations in the future. If the author does not want metadata, then he agreed to the inferior query results of PhysDocI http://physnet.uni-oldenburg.de/PhysNet/physdoc.html. The posting first, refereeing then is not the problem in the e-age, since it allows richer variety of vetting, quality filtering, peer reviewing for some, but sufficiently for effective research search engines, richer metadata and ref encoding, etc. Ebs