Here's what I would call a very welcome idea from the new editor of Economic Inquiry, Preston McAfee:
More insidious, in my view, is the gradual morphing of the referees from evaluators to anonymous co-authors. Referees request increasingly extensive revisions. Usually these represent improvements, but the process takes a lot of time and effort, and the end result is often worse owing to its committee-design. Authors, knowing referees will make them rewrite the paper, are sometimes sloppy with the submission. This feedback loop - submitting a sloppy paper since referees will require rewriting combined with a need to fix all the sloppiness - has led to our current misery. Moreover, the expectation that referees will rewrite papers, combined with sloppy submissions, makes refereeing extraordinarily unpleasant. We - the efficiency-obsessed academic discipline - have the least efficient publication process.
The system is broken.
Consequently, Economic Inquiry is starting an experiment. In this experiment, an author can submit under a 'no revisions' policy. This policy means exactly what it says: if you submit under no revisions, I (or the co-editor) will either accept or reject. What will not happen is a request for a revision.
I will ask referees: 'is it better for Economic Inquiry to publish the paper as is, versus reject it, and why or why not?' This policy returns referees to their role of evaluator. There will still be anonymous reports.
Authors who receive an acceptance would have the option of publishing without changes. If a referee noticed a minor problem and put it in the report, self-respecting authors would fix the problem. But such fixes would not be a condition of publication.
Via Tyler Cowen, on Marginal Revolution. (the post of Cowen also has already some extremely interesting comments)
Update, July 29th: A very interesting paper linked to the topic and discovered by reading the comments to Cowen's post above (and probably influencing the decision of McAfee) is Tsang and Frey's (2006) "The as-is journal review process: Let authors own their ideas". You can download here the final working version- you need a subscription to download the published one.