How scientific publishing works

With the big fiasco about Dr Hwang Woo-suk fabricating results of his cloning experiments, the BBC has a great analysis on peer review. To those outside the field, here is an explanation about peer review and how scientific publishing works.

Scientific publishing is based on the ‘peer review’ model. Essentially, this means that the scientific paper (the manuscript) is submitted to a journal, which in turn sends the manuscript to 2 or 3 reviewers in the field. These reviewers, the peers, are scientists who know the subject very well and can spot mistakes and scientific flaws. They can recommend that the paper be published, or can request further data, or better presentation of data, or doing more experiments, or, in extreme cases, outright rejecting the paper. The paper’s authors do not know who are the peers that reviewed their paper; i.e., the reviewers are anonymous.

For most papers, the reviews vote for publication after fixing a few minor errors (typos, graphs, etc). The reviewers’ feedback is then sent back along with the manuscript to the paper’s authors who get to comment on the reviewers’ comments. Usually, the feedback is accepted and the paper is submitted in final form.

The final paper is then received by the journal, which formats the paper, creates the digital copies, and publishes it in paper and/or digital format.

This system works very well, but as the BBC analysis quotes publishers as saying, it’s open for abuse. The best (and really only) way to completely check a paper’s results is to do the experiment yourself. This, however, adds a great magnitude of complexity to the publishing process. In reality, the review process just makes sure that the paper is sound: it fits in the field well, it asks good questions, the experiments are a good way to answer those questions, and the results of those experiments are interpreted well.

The scientific community at large checks the paper’s results because there are always other people interested in the field. These other groups read the paper and try to replicate the experiments and note the results. Usually, the same results are found, and all is good. If there is a discrepancy, then it could be something interesting is going on, or there was fraud. Further investigation can discern between the two latter cases.

So now that you know, go read the BBC analysis. Hopefully this helps you understand it better