Having been knee deep recently responding to both grant proposal and paper reviews (as reviewee), I find myself more and more annoyed with the review process. Sometimes, maybe even most of the time, reviews can be helpful. But we all get those off-base or nitpicky reviews that are at best a tedious annoyance or at worst, a grant killer. Anonymity and lack of public accountability for what one writes in a review, I think, gives some reviewers carte blanche to shoot from the hip, often causing collateral damage.
There's a solution: Make the reviews public.
Not that many people would read them. We've already got more than enough to read. But maybe just knowing that your off the cuff remarks might be subject to public ridicule -- on some blog, for example ;-) -- would be enough to induce a little constraint and rationality.
There are other benefits to public reviews. Sometimes reviewer-reviewee exchanges are highly instructive, and sometimes more interesting than what goes in the paper. It could be beneficial, or at least discussion-provoking, to see these behind the scenes debates. Published reviews could also cut down on work when responding to reviews: when you get the same criticism over and over again, you could just cite a previous review response rather than writing a whole new response every time ("see Hickok & Poeppel review response 2000, 2004, 2007 for repeated and thorough dismantling of the same tired point you raise here"). It might also promote more constructive criticism in reviews, or even more willingness to review papers because the reviewer would get some credit for suggesting that clever control or theoretical insight. Folks might even become so proud of their reviews that they might start signing them and listing them as pubs.
Maybe we'll start publishing reviews of our papers here. I wonder if that would cause a stir.