This cart-and-horse problem is all known, of course, and was the main reason that PLoS was such a great idea--a sort of jump-start. However, most of us aren't (yet!) putting out Nature- and Science- and Cell-level papers, and I think the stubbornness of this problem for the good-but-not-world-class journal level has been grossly underestimated. It all comes back to that limited monopoly crunch. Lots of journals at the bottom. A handful of journals at the top. Still only a few just below the top in each field. PLoS solved the very top problem handily, or is making great strides. But the Biol. Psych. dilemma will be with us for some time. New journals start at the bottom and have to rise, unless they are jumpstarted like PLoS. Part of the answer is to keep on going, accept that change will come gradually, and, like PLoS is doing, have them add their impramateur to new journals in the family. Another part is to convert for-profit journals to OA.
But PLoS can only start so many journals, and for-profits have tended to resist OA. The problem will be with us for a long time.
Today I locked up my paper with a signature.