"Rejecta Mathematica, which is online only and open access, has some simple rules for publication. Papers had to be submitted for peer review at one or more other journals, and they had to have been rejected, either by the journal's editors or the paper's reviewers. The papers authors could then send a letter that describes the reason for rejection, along with the paper itself, to Rejecta, which would publish both as a single document."
I've always wanted a Journal for Irreproducible Results or something. This isn't exactly that but the intention is there, and it's a (gnat's) bite in the butt of big sci-publishing. Alas, the implementation may seem a bit too permissive.