I found this passage from an email asking me to referee a paper very perceptive. The editor wrote:
From personal experience, there is a (growing?) tendency for referees to provide very long, detailed reports that sometimes read - with slight exaggeration - like an essay on the question “how I would have written this paper”. This is not always optimal - it costs the referee’s time, and it may result in a lengthy, non-converging refereeing process overall. Our main interest is in receiving advice on a. the overall significance of the findings and claims b. their technical correctness. In my experience, a relatively brief report containing a set of consecutively numbered paragraphs, each making one point, is most useful in reaching a final decision.
Update: This quotation received an extraordinarily positive reaction on my Facebook page (with 201 likes and counting, including many from well-known economists, plus several positive comments), as you can see here.