The Scientist on peer review
Sometimes peer review is stellar and insightful—there are peer reviewers I wish I could have given co-authorship or sent flowers to!—and sometimes quixotic. More fundamentally, peer review is inherently limited, unless one has access to underlying raw data, and some greater sense of the lineage of scientific choices than is reflected in a manuscript. Think back on the world-shaking Hwang claims a few years ago that somatic cell nuclear transfer had been achieved in human beings, and the review finally required to establish falsification and to uncover the ethical issues with human egg donations.
I believe that without fundamental changes in data access, like widely pooled data, and iterative collective examination of aims and methods, peer review can never do all we ask it to do, which is to compare and validate approaches, data, results, and directions, in themselves and as stepping stones to further scientific work and informed policy-making.
One notes that there were internet reports, emanating from South Korea early on in the game, suggesting the falsity of the results of Hwang Woo Suk. The journal Science initially referred to the problems as a mere mixup in photographs. Although there was indeed much review, not so much review was needed to establish the falsity of the results.
An article in JPTOS discussed some aspects of Science's publishing of the fraudulent work by Hwang Woo Suk (Analyzing Innovation the Right Way, 88 JPTOS 239).
See also Who Can You Believe In Academic Writing?