If I may add my two cents to this discussion, what is needed is not a revision of this or that definition but an unequivocal legal license by which the author permits the perpetual distribution of his/her work and of derivative works on the Internet. Immediately and irrevocably.
Next thing needed is the technological realization of a universal open access library on the Internet, where everyone can distribute the works one finds worth of investing the corresponding resources. There is no such thing as a permanent access to anything. It only exists as far as there are interested parties in preserving the access. The technology should make it easy for anyone to add his or her resources to preserve any given publication. I believe that the OAI protocol is a very good first step in this direction, but it is not at all completely satisfactory. Third and most difficult of all, what is needed is the articulation of a social movement by which authors get sufficiently motivated to give top priority to the use of said license (or one of a set of said licenses) in publishing their works. This movement should be strong enough to convince the publishers, for profit companies, not for profit university publishers or scientific societies to accept processing and printing the papers under such a license and without the authors transferring their copyrights integrally. Eventually, an exclusive right to a printed edition may be part of said license. Of course, something like this is already done by some OA journals but the community needs to extend this to all publishers! This might sound very far fetched but I would like to call your attention that some authors are making progress along these legal lines and I believe that a license like that would be essential at this time for the OA movement. I just wish the lawyers could soon come up with something capable of obtaining the enthusiasm and compromise of our academic communities. Such a license should be a social contract clearly regulating the rights and obligations of all the interested parties in the publication process: authors, employers, founders, editors, peer-reviewers, readers, etc. I wish to finish with two examples I alluded to before. Larry Lessig published his last book, Free Culture, <http://www.free-culture.cc/> under a Creative Commons license which allows unlimited non-commercial distribution of derivative works. Yochai Benkler published his seminal paper "Coase's Penguin, or Linux and the Nature of the Firm" <http://www.benkler.org/CoasesPenguin.html> in the Yale Law Journal, retaining the copyright. Concomitantly he published his paper on the Internet using a Creative Commons license. For more details, see the following errata: http://static.highbeam.com/y/yalelawjournal/june012003/erratacorrectionnotice/ Copyright notice reading, "Copyright [c] 2002 by The Yale Law Journal Company, Inc.," should read, "Yochai Benkler, Coase's Penguin, or, Linux and The Nature of the Firm, 112 YALE L.J. 369 (2002) copyright [c] 2002 by Yochai Benkler. All other pieces herein copyright [c] 2002 by The Yale Law Journal Company, Inc." Just my two cents. Of course, a zillion details are missing from this scheme and should be filled-in in due time. But my third cent, if I may be allowed, is that it is not only worth working along these lines while building up OA but indeed, these lines are essential for the success and widespread adoption of OA by the academic community. Cheers, Imre Simon