Here are three [sic] problems with peer review:
(1) It’s a (mild) hassle to submit a paper, requiring the navigation of (mostly) poorly designed online systems, the writing of otiose cover letters, and so on. Since most papers don’t get accepted at the first place they’re sent, this mild hassle is multiplied several times.
(2) Once the editor has received a paper, they are deprived of a very important piece of information about the paper, namely whether and by whom it’s been reviewed before and, if so, what the reviewers thought and how the author responded.
(3) A referee’s work is often in vain: people will submit, a referee will work hard to point out problems with the paper, the paper will get rejected, and the author will submit again elsewhere relatively confident that the referee will not review it again.
(4) It is hard to find referees.
I think a system aimed at solving (1), the least serious problem, could help with the other three problems: in brief, the proposal is a journal-neutral submission system (call it JNSS), a single front-end from which authors can submit and resubmit a given paper to different journals without having to navigate individual sites or redundantly enter the same information numerous times. Onto this front-end an editorial backside could be added, that contains information about the paper’s history and is accessible to only and all journal editors, enabling them to see who has already reviewed the paper and what those reviewers thought of it.
Readers should check out the rest of McKeever's post to see how he thinks this might work. For my part, I think a substantial amount of McKeever's proposal might be a good idea. For instance, I've heard (though I may be wrong) that law journals use something like his journal-neutral submission system--which I think might make things easier on authors, editors, and referees. I also like McKeever's suggestion that authors who submit papers should have to agree to be included in a pool of referees--as a central pool (categorized by topics and author AOS) might make it much easier for editors to find reviewers.
However, I am much less convinced that it would be good to forward past referee reports to editors at new journals, or to give editors at journals inside information about journals a paper has already been reviewed at. My worry is that this might bias editors against papers that have been bouncing around a while--papers that may well be very good. I don't think we should hold a paper's past against it--particularly given that its author may substantially revise and improve it. For these reasons, I'm inclined to think every new paper submission should be treated as a "clean slate."
But these are just my thoughts. What are yours?