The current system of having authors ask for authorship and demonstrate their authorship is both unnecessary and annoying. It is unnecessary, as papers hardly ever have authors that cannot defend the work (despite what you read on xkcd), and it is cumbersome, because right now you can't find the attribution of the papers, nor do we have cumulative author scores on authors who aren't here yet.
I think it would be good for the site to assign the author based on the publication authorship list without waiting, and to create the author profiles automatically for all physicists whose work is reviewed, and then do a manual merge of those authors that use several different forms of their name.
We would like to allow anyone to import papers to reviews soon, not just admins, and this would make the task easier--- it would also assign the proper reputation to the proper author.
It also can lead more interest in the site, as we will give Weinberg a score (currently 4), Witten a score, and so on, and authors will want to know how they are faring reputation-wise, just out of curiousity, like "rate my professor" or "hot or not", even if they do not contribute directly. This can lead to a lot of eyeballs and votes. To start the process off, you can import some classic papers by Einstein, Weinberg, and so on, and give them some upvotes.
For a more serious example, I asked for Marco Frasca's paper 0709.2042 to be imported, as a paper sorely demanding a lot of hostile peer review. I think it would be good to associate this paper with his existing account, and he will not be here to do it.