Since some time there has been discussions about having a place for community reviews; see for example
Such a system already working at PhysicsOverflow with considerable success.
My attempt to start something similar in MO was quickly rejected for reasons that I well understand: MO is a place for questions and not for reviews. Therefore I intend to open an area 51 site.
I would keep it simple: Users (anyone, not only authors) can post links to freely available preprints together with some info (authors and abstract). The only acceptable answers would be classical journal-type reviews (not mathscinet type reviews). Comments should only address points that need further clarification. The site should not a place for discussing the paper's impact. Anonymous reviews are of course accepted. The flagging and point system should work well enough for "accepting" or "rejecting" preprints/reviews. Sectioning could be done via AMS subject classification.
But for all these I need at least 5 users to support the site and 5 example questions. I am willing to put my 3 current unpublished preprints as questions.
It would be great to have a discussion, not just upvotes/downvotes. If you support this, please let me know. If you think that such an effort is wrong, please state your argument.
EDIT1: It could be considerable having a list of members/moderators acting as "editors", that is, they would receive preprints from authors that are not members of the site, post their preprints and recruit reviewers. I am willing to act as an editor for pdes and calculus of variations.
EDIT2: Following Gerhard Paseman's suggestion, it makes sense to ask those that want to have a paper reviewed, to provide with at least 3 classes for which they are willing to act as possible reviewers. I am willing to review for: 35A01/2 (PDE - Existence/Uniqueness problems) 35A15 (PDE - Variational methods) 35Bxx (PDE - Qualitative properties of solutions) 35Gxx (PDE - General higher-order equations and systems) 49Jxx (Calculus of variations - Existence theories)
EDIT3: In the review section of PO (3) The review section, a mechanism is described how manuscripts, may be accepted for review. We should also do something similar here in order to avoid submissions that contain work-in-progress whose authors expect to get community advice. For example let the question reach 10 upvotes (including 1 from an subject editor) before it is eligible for review.
EDIT4: Just to be clear, I am willing to write down the first 5 reviews needed to keep things going.