This is to add to Stevan's very thoughtful reminiscences some of our own experiences in physics concerning relations with publishers and OA.
In 1994 there were several workshops with participants from different countries, organized by either Frank Laloe (CNRS, Univ. Paris Sud) or myself in Oldenburg. Bob Kelly from APS and H.E.Roosendaal, the strategic chief person from Elsevier Amsterdam, attended. We got an invitation from Elsevier to Amsterdam and reported to the Chief executive committee on the top floor. Beforehand, when they saw me from the windows entering the building down below, someone quipped "Here comes the death of Elsevier"; however, it was a very constructive open discussion. We had a committee -- 'Elfikom': electronic research information and communication -- of which Springer was a member. We had joint policy papers by Roosendaal (ES), Springer, IoPP, and ourselves, on the future of distributed document databases, integrating the physics self-archiving system PhysNethttp:// www.physnet.net with joint retrieval, etc. We had a joint European Union Application DDD-Physics 'Multimedia scientific Physics Document Database in Physics' with what was (in retrospect) an impressive list of participants (CERN, Physics societies, user groups, FIZ Karlsruhe (STN Database host), Rank Xerox, Akademie Verlag, Elsevier, Springer Verlag in 1995. The link is of course still there, although some links from there point to nowhere now: http://www.physik.uni-oldenburg.de/ddd-phys/partners2.html http://www.physik.uni-oldenburg.de/ddd-phys/ the homepage http://www.physik.uni-oldenburg.de/ddd-phys/proposal_small/sectionstar3_3.html where you still can read the whole application. (I will look for the password if asled.) There were constructive discussions with APS, and IoPP. The Action Committee for Publication and Scientific Communication of the EPS had publisher members (IoPP and EDP). But then it became clear that to take this approach was to drag our feet. EPS remodeled its Action committee, no publisher members any more, DDD was turned down by the EU, and PhysNet became autonomous. Only now (for the past 1-2 years) can the retrieval engine include the abstracts of all IoPP papers, thanks to a constructive agreement. The elfikom http://elfikom.physik.uni-oldenburg.de/ server is still there, started in 1995, and last changed in 2000. Publisher Members were from Springer, Elsevier, Akademie Verlag, VCH, Phys. Blaetter http://elfikom.physik.uni-oldenburg.de/Docs/Mitglieder_english.html A list of early activities of 1994/1995 was compiled in 1998: http://elfikom.physik.uni-oldenburg.de/Docs/Termine.html The talks of 1994/1995 are also still there. http://www.isn-oldenburg.de/~hilf/vortraege/halle-ebs/halle-ebs.html http://www.isn-oldenburg.de/~hilf/vortraege/bmftprojekte.html which recur to older activities. And the papers, http://www.isn-oldenburg.de/publikationen.html?pub=liste The IuK www.iuk-initiative.org Initiative Information and Communication of the Learned Societies in Germany was founded in 1995. Thus I conclude that the dissociation from the publishers occurred later, in late 1998, when they began to leave in order to hold on to what they had, and now they fight even harder to maintain toll access as long as possible. Elsevier is going to court in Germany against the University Libraries concerning document copy delivery. The Government intends to forbid document copy delivery in future, allowing documents to be viewed only on-screen in the library. See http://www.urheberrechtsbuendnis.de I still think that closer contact and bridge between publishers and research University groups would have boosted technological development much more than the present dissociation has done, and thus it would have served the science process more. This would have meant focussing on additional, new, innovative professional services and letting document ownership (now called toll access) become free by embracing OA throughout. Eberhard Hilf