The panel of reviewers for peer review is selected informally by the editor. Typically, the editor will have access to a large pool of reviewers and will randomly assign three to five to review a particular paper. One important part of the process is that the reviewers are considered the authors "peers", and are not considered more knowledgeable, more accomplished, or of higher social status than the author. In this respect, peer review differs sharply from forms of evaluation which occur in an educational or business context.
The process of peer review in a scientific journal involves the editor of a journal sending an article to several peer reviewers who remain anonymous. The peer reviewers evaluate the article and add comments for improvement. These comments are sent to the editor who then transmits the comments to the author.
Some of the characteristics of the systems are:
Sociologists of science have analysed the way in which the system of peer review works in practice and have advanced the criticism that it is vulnerable to capture by an elite. Experiments have been performed which suggest that peer reviewers tend to report, or abstain from reporting, logically identical errors in papers submitted for peer review accordingly as the paper's conclusions contradict, or substantiate, the conclusions of the reviewer's own research.
This tradition has long been known to the academic community[?], especially with scientific papers, where publication of the paper allows its quality and veracity to be criticised, the purported goal being to improve the quality of work.
In the field of computer software, the principle mentioned above has been stated as Linus's law, often formulated as "given enough eyeballs, all bugs are shallow". Eric S. Raymond has also written extensively about peer review as it pertains to open source software, in a series of papers starting with The Cathedral and the Bazaar.
Having one's work criticised and improved is intended to have the beneficial effect on improving the quality of their future work.
See also: preprint