The second one is the hard one, because prestige is the very thing they have.

Something like PageRank (itself a citation-weighting scheme) should do it, based not on the coarse-grained prestige of editor-gatekeeper publishers, but the fine-grained prestige of researcher-authors and the work itself.