Hidde Plough writes an incisive critique of the current system of publishing in "top" scientific journals, emphatically entitled "End the wasteful tyranny of reviewer experiments." The agreement on its deficiencies among scientists, myself included, is unanimous.
I think that we should go beyond this critique, and move scientific publishing towards the PLoS ONE model-- with publication based on technical validity only-- but that would deprive for-profit publishing companies and academic hierarchies from using publication in "top" journals as business instruments and measures of scientific productivity/fundamental contributions, respectively.
The whole goal of "top" (read, general interest) journals of providing a general interest forum to the readership is obsolete. Today, electronic papers can be tagged, annotated, and searched regardless where in the library stacks their bound volumes may be located. Physicists have been doing this for years at arxiv.org with some of the most important work being deposited there years before peer review.
But for now, as Hidde concludes, editors and reviewers should resume their true responsibilities, and "revisit the experiment."