Good post. The way I see it, journals have traditionally provided three core features: publishing, quality control (via peer review), and reputation (via exclusivity). The publishing aspect is basically solved -- just post your article to the arXiv.

The obstacle is the other two -- quality control and (in particular) reputation. What we need, and what you've more or less described, is an independent peer-review/evaluation platform where contributors can build & show-off reputation.

The good news is there is a lot of activity around this now. My startup Publons.com is one of many ventures trying things in this area: PubPeer, Peerage of Science, The Winnower, Libreapp.org, peerevaluation.org...