[thaw13579]

Aside from the question of being lawful, I don't think they quite fit the bill alone for a few other reasons. First, they are not truly public or accountable to the scientific community -- while they accept uploads, they are both designed and managed by unaccountable private parties. Second, they do not maintain much in the way of metadata, making it hard to verify basic details such as the publication date, authorship, and revision history.

This second points is important for having a coherent discussion of the literature and avoiding fraud to some degree, and to their credit, the major journals do provide this (but not the first point). In this sense, I think Pubmed Central is an example of a good way towards a public utility model:

https://www.ncbi.nlm.nih.gov/pmc/