you address an important part of scientific publishing, and I agree with what you have presented there. However, there are (at least) two problems that I find cropping up time and again with virtually all beyond-traditional-publishing initiatives: weeding and geekyness.
Weeding the huge amount of unreliable, post hoc speculative, incompetent, uninformed or simply unscientific "contributions" to science is a non-trivial task. Following MacArthur's dictum of "You can either keep up with or contribute to the scientific literature.", I simply do not have the time to follow blogs and twitter and about a dozen pre-print outlets and early-views and "traditional" eTOCs and F1000 and so forth, PLUS then giving out points for what I perceive as "good" articles. Without pressing the virtual "Like it"-button, however, no weeding is achieved, and the pile of new (and largely unreflected and poorly crafted) contributions grows faster and faster.
Geekyness is the technical problem attached to weeding. I might be able to programme in C++, but that does not make me wanting to set up a "working" environment that constantly flashes new information at me. I cannot be asked to have SVN and github installed, LaTeX and Markdown, maintain a blog and be forced to use specific formats when submitting to a traditional journal.
I find the rigor of current ecological science wanting, much more than its dissemination formats.
Best wishes, Carsten