Three myths about scientific peer review
michaelnielsen.org
michaelnielsen.org
As a peer reviewer, I've sent papers back with comments along the lines of "this work seems quite similar to the published work of X in the following papers (which the author doesn't cite): ... I wonder if he if the author is aware of this earlier work", and subsequently heard nothing more (presumably because, upon looking at said papers, the author came to the conclusion that he had not in fact discovered anything new). On the opposite side of the coin, one of my papers benefited from very similar comments, in that having related work brought to my attention allowed me to revise my paper to better describe how it fit with the rest of the field.
I think it's in cases like this -- where peer reviewers point out things (whether other research, or possible errors) of which the author(s) do not seem to be aware -- that peer review really shines. If reviewers respond merely with a 'yea' or 'nay' they're not doing their job properly; their job is to review, not to judge.
I don't think that's what the author concludes. But otherwise, I agree with you.
PRL, not so much. In my experience, at least one referee doesn't even finish reading the paper. It's not just me, this happened to a friend of mine last week:
Referee: "The authors do not demonstrate that their equations apply to any physical example."
Response: "See page 3 paragraph 4."
I do think the peer review system will eventually change to something more like hacker news. It is a much more convenient way to do things than trying to find volunteers, mail out papers, collect responses, etc. Not to mention you can potentially draw from a larger audience and you make your work more easily accessible.
One mechanism that does exist in peer review that I'm not sure could be easily reproduced is a sense of duty while reviewing a paper. When selected to review a paper, you know you are one of only a few who will determine the fate of this paper and you feel a responsibility to try hard to understand the paper and give it a fair critique. If a paper were just posted online for anyone to critique, people might not put forth as much effort in their reviews.
How about completely open reviewing, like in some of the BiomedCentral medical journals (e.g., http://www.biomedcentral.com/bmcmed)?
If done properly, all the cards are on the table. A paper cannot (should not) be rejected or accepted without legitimate reasons. Reviewers and authors can be made to account (through tarnished reputation and risk of embarrassment, at least) for their actions.